CHANGELOG.md

Change Log

All notable changes to this project will be documented in this file.

1.2.9 - 2024-3-12

Added

• If you would like to have the namespaces sorted in a (:require ...) in an ns macro, you can use the style :sort-require. You can also select (using regular expressions) namespaces to put first (or last, or both). Search for :sort-require in the reference manual for details.
  This will also, by default, sort the identifiers in the :refer vector, though you can disable that sorting.

• If you configure :comment {:min-space-after-semi 1} then a comment like this: ;this is a comment will be changed after formatting into a comment like this: ; this is a comment. The default for :min-space-after-semi is 0, so by default no changes are made. Issue #306.

• New option map to transform random Java classes into other, more manageable data types: :modify-sexpr-by-type. Also allows options map to be used when particular types are encountered, and enhanced color map handling supports specification of colors when using :new-l-str and :new-r-str in option map by using the color map values of :left and :right. Issue #307.

Changed

• The style :ns-justify was difficult to change in situations where it wouldn't justify something because the maximum variance was exceeded. It has been restructured using the new style-map approach to allow it to be easily parameterized. If you invoke :ns-justify by using its keyword name, the maximum variance for justification will be 20 for :require and :require-macros and will be 1000 for :import. You can change any or all of these by using a map to invoke the :ns-justify style. Invoke it with {:style-call :ns-justify :require-max-variance 1000} to essentially force it to try as hard as it can to justify the :require list in an ns macro. Likewise, you can use :require-macros-max-variance and :import-max-variance to adjust the maximum variance for these sections of the ns macro.

• zprint will now recognize when a set of pairs is being spliced in using reader-conditionals, and format them as a set of pairs. This comes up when splicing in a series of local symbols and expressions into a let, or splicing in a series of key value pairs into a map. If there are an even number of elements in the reader-conditional and it appears in a place that expects pairs, then the insides of the reader-conditional is formatted as pairs. NOTE: this can (rarely) result in unanticipated changes in formatting where there was no source change. In the event that happens to you, if the new version isn't "better" than the old version, please submit an issue. There were two small changes in the formatting for zprint, and both were considerably better than the previous formatting.

Fixed

• Fixed two reflection issues. Issue #305.

• Extended the meaning of :one-line-ok? for {:style :rod-config} so that if a particular arity would format successfully onto one line, it will now do so if :one-line-ok? is true. Issue #282.

• There was a bug where if you specified multiple styles in a vector, if there were any errors detected when applying the styles, the errors were lost unless they occurred in the rightmost (that is, final) style. Any styles that had errors wouldn't be applied, but there would be no indication of that and it would all appear to be successful. This has now been fixed to detect and report errors from any styles where there was a failure to apply the style. There is a small chance that this will appear to break some working configurations, since if there were errors applying multiple styles some of those errors would not currently be reported if they were not associated with the final style. In any situations where these errors suddenly appear, it signals errors that were not properly exposed to the user now being reported. You can restore the previous behavior by simply removing the style that causes an error, since it was not being applied previously. You may instead want to figure out how to get the benefit of actually applying that style by correcting the error, since that was probably the original intent. Issue #311.

• The style :minimal-smart-wrap added in 1.2.5 had a problem where it wouldn't properly deal with single lines ending in a period. It also didn't recognize lines ending in a period if it wasn't preceded by two lower-case characters. Both of these problems are now fixed. Issue #297.

1.2.8 - 2023-09-14

Added

• New capability -- remove an entire part of the options map with :remove-final-keys [[ ... ] [ ... ]]. Used in resolving Issue #301, where someone wanted :indent-only kind of behavior, only with justification as well for just a few functions. To do this, you need to disable essentially the entire :fn-map, and then put in place only a very small one. Thus :remove-final-keys [[:fn-map]] will remove the default :fn-map. Then you must put a new :fn-map in place iin the same options map, or zprint will not function. This is a rarely needed capability, but a hard one to live without if you really need it!
• New pre-compiled binary for macOS running on Apple Silicon. Install zprintma-1.2.8 from the release to get it. Runs up to 3x the speed of the Intel binary on Apple Silicon! Discussioin #302.
• Added support for using babashka bbin to easily install zprint.

Changed

• Modified the library to run in parallel when using babashka.
• Replaced the old test_config bash file with a new babashka capability to do the same thing, only better. bb testconfig:bb -v 1.2.8 -s uberjar will run it.

Fixed

• When using babashka, you can now use functions in the options map in any configuration file. This is possible due to changes in babashka that exposed sci so that zprint could use the same code in babashka that it uses in :clj.

1.2.7 - 2023-06-08

Changed

• Enhanced :style :minimal-smart-wrap to never flow words back up onto a line terminated by several lowercase letters and a period.

• Enhanced {:comment {:smart-wrap? true}} (the default) to not touch any lines starting with (, {, [, or ending with ), }, ]. This will mess less with code that has been commented out. Try to avoid actually commenting out code, use #_ instead in front of the expression to ignore. Much cleaner in several dimensions.

Fixed

• Top level comments (those not contained in an expression) were wrapped using the {:comment {:smart-wrap {:border n}}} instead of the {:comment {:border n}} border.

1.2.6 - 2023-05-27

Added

• You can now specify some keys to come last in a map as well as some keys to appear first in a sorted map. The {:map {:key-order [...]}} configuration places all of the keys prior to the distinguished key :| at the front of the map, and all of the keys after the :| key at the end of the map. If you have a key :|, you cannot specify its order in the :key-order vector. If this is a problem for you, submit an issue and I'll make it configurable. Issue #281.

• Added function type :list, for when you want a list interpreted as just a list of data, without assuming that the first element of the list is a function of some type. In addition, :nl-separator? has been also added to :list in the option map. It only has meaning when using the :list function type. In that case, if it is true, every list element that requires more than one line will have an additional newline after it, resulting in a blank line. In this case (where :nl-separator? true), :nl-count will be used as the number of newlines to use after some element that takes more than one line. The default is 2, but you can set it to be anything you wish. In this situation, a vector value for :nl-count is not allowed, and if a vector is configured, the value 2 is used instead. Issue #283.

• If you were using zprint as a library, and you kept the configuration you wanted to use with zprint in a file of your own, there was no safe way to give that configuration to zprint if it contained function definitions -- which it might need to for :option-fn values. Now you can pass a string value of an options map to set-options!, and it will read and 'compile' that options map (including any functions) using the Small Clojure Interpreter (sci) built into zprint, the same way that zprint handles reading external configuration files. Issue #283.

• Configurable Styles: As the number of styles that call :option-fn to return guides or just to do complex things has grown, many of these styles have also begun to accept a configuration map as their first argument. This has worked well, but has in turn required a new style in the :style-map for every unique combination of values in the configuration map for the option-fn (or has required moderately complex configuration in the :fn-map or elsewhere). The processing for styles has been enhanced to create a :style-fn, which is a key which appears in a style in the :style-map. It is called with sufficient information to allow it to return a style (possibly containing a configuration map for an option-fn). A :style-map style containing :style-fn can be used just like any style, but it can also be used by giving a map to :style, where that map contains a :style-call key whose value is the style in the :style-map. The map with a :style-call key can also contain configuration information for the resulting option-fn. Thus, you can invoke a style and pass it configuration information when you invoke it, without requiring you to define yet another new style which encapsulates your particular configuration. See the style :rod-config in the :style-map for an example of how this might work, and the documentation for style :rod in the reference manual for additional details. This was added in part because of Issue #282.

Changed

• The handling of tagged literals (for example, #js) has been enhanced. By default, zprint will try to put the next thing on the same line as the tagged literal. There is now a configuration element in the option map: :tagged-literal which you can use to configure zprint to operate the way that it used to by default -- {:tagged-literal {:hang? false}}. Issue #292.

• Comment handling has been considerably altered. When working on the stability fixes for 1.2.5, the largest remaining problem was comment wrapping causing changes to the formatting in subsequent runs. In addition, the comment wrapping has been very simplistic since its inception, leaving wrapped comments looking pretty bad. There is a new capability called {:comment {:smart-wrap? true}} which will now word wrap comments cleanly. It will also repair most of the problems that the simplistic wrapping produced in the past. It is now the default, in no small part to repair the problems of the past. If you are working to minimize changes when running zprint, I would recommend running it once over your code before you disable it, as will clean up most of the problems that were added by zprint in the past. You can disable it with {:comment {:smart-wrap? false}}. You can also configure it to minimize the amount of word wrapping it does, while still allowing it to do much better than the previous default by using {:style :minimal-smart-wrap}. You need to have {:comment {:smart-wrap? true}} to use :minimal-smart-wrap. See the reference manual for details on how to configure smart wrap. Issue #271.

• Upgraded to [org.babashka/sci "0.7.39"] and [rewrite-clj/rewrite-clj "1.1.47"].

Fixed

• The style :hiccup was sometimes recognizing destructuring in argument vectors as a hiccup vector. Now it doesn't do that as much, though it is still possible. Issue #294.

• Found and fixed several areas of justification that were slower than they needed to be. Also added a new configuration element, :max-depth to every :justify map {... :justify {:max-depth n}}}. Justification will only be performed until the :max-depth is reached, and then it will not be attempted. This is only necessary when justification is desired and there are very deeply nested expressions where there are justifiable elements (e.g., maps) deeply nested in expressions that don't fit well within the width. Issue #290.

1.2.5 - 2023-01-24

Added

• Prior to this release, when a pair was formatted, if the left-hand-side was a multi-line collection, the right-hand-side would always flow. That remains the default, but if you enable :multi-lhs-hang? true in :binding, :pair or :map, the right-hand-side will format on the same line as the last line of the multi-line collection on the left. Issue #273.

• In the event that you are using justification, and you have enabled :multi-lhs-hang? true in :binding, :pair, or :map, the justification code will attempt to "squeeze" collections on the left-hand-side when performing justification. It will attempt to squeeze them into some fraction of the remaining space on the line, with the fraction configurable as :justify {:lhs-narrow 2.0}. The space into which zprint will try to format the collection on the left-hand-side is controlled by :lhs-narrow, which is the denominator of the fraction of the remaining width in which it will try to format the left-hand-side. The default for :lhs-narrow is 2.0, which means that it will by default try to fit the a collection on the left-hand-side of a set of pairs into 1/2 of the remaining space on the line. :lhs-narrow is only used when justifying and :multi-lhs-hang? true is enabled. Issue #273.

• In order to prevent an argument list from having something hanging out on the right end (e.g., a "&"), you can now use :vector {:no-wrap-after #{"&"}}. You can have any strings in the set of elements in :no-wrap-after. Issue #252.

• A couple of new capabilities. Several people have wanted regular expression capabilities in the :fn-map. A couple of people have also wanted a way to format functions where the function name was over some length differently than they are usually formatted. Both capabilities are now available by using a built-in :option-fn. See :rule-example (and :regex-example) in the :style-map for some worked examples. The :option-fn rulesfn accepts a vector of pairs of elements as its first argument. If the left-hand-side of a pair is a function, it will call that function with the string format of the function name. If the function returns non-nil, then the right-hand-side of the pair is returned as the option map. If the left-hand-side of the pair is not a function, it is assumed to be a regular expression, and it is matched against the string format of the function name. If it matches, then the right-hand-side of the pair is returned as the option map. The pairs in the vector are processed in order until one of them has the right-hand-side returned as an option map or they are all completed, in which case the :option-fn returns nil. Issue #261 and some others.

• When an anonymous fn is compiled, alll #(...) forms are turned into (fn* ...) forms. If you then format the compiled structure, it looks pretty bad. If, instead, when you format the compiled structure you use :style :anon-fn, zprint will backtranslate the stucture to reconstitute the #(...) forms. Issue #268

Changed

• Multiple aligned inline comments shift left to be one space beyond the widest line of code when using :comment {:inline-align-style :aligned} or :comment {:inline-align-style :consecutive}. Prior to this release, single line inline comments would not shift left. Now all inline comments will shift left to be one space beyond the widest line of code, unless :comment {:inline-align-style :none} is used. Issue #271.

• When justifying, different tuning is used for decisions about when to hang or flow the right-hand-side of a pair than is used when not performing justification. This tuning used to also be used for the formatting of the right-hand-side of the pair, but is now independent of the formatting of the right-hand-side pair itself. This causes subtle changes (typcially improvements) in the formatting of justified pairs.

Fixed

• Some files changed the second time they were formatted, particularly when using :style :respect-nl. Generally they don't change when formatted a second time now, both with the default options and with :style :respect-nl. The most common reason for a file changing in a subsequent format now is that a comment exceeded the :width on the first format, and the comment was wrapped -- adding another line. Sometimes this additional line can trigger a change in a formatting heuristic which can change the formatting in a way that is visible. The changes to resolve this issue caused a number of subtle formatting changes in many places, though nothing particularly unpleasant. If you wish to forego the repeatability changes and replicate the formatting of previous verions as much as possible, you can use :style :original-tuning. Issue #271.

• Empty vectors removed when using :format :off. Issue #263.

• ->> incorrectly formatted when using :style :community. Issue #262.

• Fixes for Issue #254, #255, #256 and #257 as part of ongoing work on large new style.

• The new aliasing capability in the :fn-map doesn't work for workspaced functions. Issue #276.

• Several bugs with advanced map options with comments in the map. Issue #269

• Namespaced maps sometimes didn't indent properly. Issue #274.

• When using the zprint comment API, :format :skip and :format :off would sometimes remove empty lists or vectors. Issue #275.

• :style :rod-no-ma-nl would sometimes fail. Issue #283.

• The version of Babashka in the field when zprint 1.2.4 was released would not work with that version of the zprint library. You needed a pre-relese. Since then Babashka has been upgraded, and now works with zprint. Issue #253.

1.2.4 - 2022-08-02

Added

• You can now alias one funtion in the :fn-map to another. The key is (as always) the function name as a string, and the value is another function name, also as a string. Thus {:fn-map {"mydefn" "defn"}} will cause the function "mydefn" to be formatted however "defn" is formatted. Issue #234.

• {:list {:nl-count <n>}} to add additional newlines in lists. (Part of Issue/Discussion #236)

• Hiccup and HTML output now available from all library fns that end with -str, and the prebuilt binaries. Use {:output {:format :hiccup}} or {:output {:format :html}} to get this type of output. This is a new, capability, ALPHA quality for now. Note that you can adjust the paragraph output in the HTML. The current default is {:output {:paragraph {:style "font-size:20px;font-family: Lucidia Concole, Courier, monospace"}}}. Issue/Discussion #214.

• Another way to prevent justification of things that will look bad. In addition to {:pair {:justify {:max-variance 20}}} there is now the, perhaps easier to understand, {:pair {:justify {:max-gap n}}}, where n is maximum number spaces which will show up between the left and right columns of a bunch of pairs being justified. Note that the :max-gap is calculated after the :no-justify and :ignore-for-variance elements are removed. So if you want them included when calculating the :max-gap, you need to adjust these parameters as well. Issue #239.

• There are pairs in functions like cond, in maps, and in bindings. When the right hand part of the pair doesn't fit, it "flows" onto the next line. A new feature is now available to flow all of the right hand pairs onto the next line if any of them flow because they don't fit. This will certainly make things take more vertical space, but the additional consistency may make the pairs more easily distinguishable. You can enable this in any of the three possible places by setting :flow-all-if-any? true in :pair, :map, or :binding. Issue #235.

• EXPERIMENTAL: Every function type is now looked up in the :fn-type-map. The value of the function type is the same as the value of the keys in the :fn-map -- either a function type, or a vector containing a function type and an options map. In most cases you don't want one function type to simply alias another, so the usual value of a key in the :fn-type-map would be a vector containing a function type and an options map to be used when this function type is encountered. Note that this is EXPERIMENTAL and may well be removed. Note also that you cannot define your own function types -- the only allowed keys in the :fn-type-map are the existing, known, function types.

Changed

• Upgraded to newer rewrite-clj, and changed edn* to of-node* to remove deprecation warnings. Issue #247.

Fixed

• Namespaced maps don't work with :style :indent-only: the namespace goes missing! Issue #242.

• Indent on anonymous fns is wrong with :style :indent-only. Issue #243.

• {:meta {:split? true}} doesn't properly handle multiple ^:stuff metadata elements. It leaves someout. Issue #245.

1.2.3 - 2022-04-13

Added

• New :range configuration options: :use-previous-!zprint? and :continue-after-!zprint-error?. These allow someone doing :range formatting on a file (string) which has ;!zprint formatting directives to have the lines in the range formatted the same when doing just the range or the whole file. :use-previous-!zprint? if true will scan all previous ;!zprint directives prior to formatting the range. :continue-after-!zprint-error? if true will ignore errors in any ;!zprint directives, and output any errors in the map returned by :output {:range? true}.

• An alternative way to format metadata was created. It is :style :meta-alt. It is presently experiemental. Issue #224.

Changed

• The :fn function type didn't handle multi-arity functions well. Now they flow instead of hanging.

Fixed

• Removed obsolete underscore from docs and specs. Issue #220.

• letfn didn't indent its fns like fns. Now it does. And if they are multi-arity, the various arities don't hang, but rather flow. Issue #221.

• Several tests failed when running with new version of planck, and therefore new version of Clojurescript. Now they don't. Issue #222.

• The styles :community and :rod didn't compose, because :guided was not considered a "body" function type. Now they do. Issue #223.

• zprint-file-str was not thread-safe, for a variety of reasons. Now it is, and configuration is thread-safe as well. Issue #226.

• A function with :binding function type would cause an NPE if it had no binding vector. That is, (let) would not format successfully. Now it is fine. Issue #230.

• Figured out Figwheel issues. Rebel-readline uses rewrite-cljs, which conflicts with rewrite-clj V1. To use rebel-readline: com.bhauman/rebel-readline-cljs {:mvn/version "0.1.4" :exclusions [rewrite-cljs/rewrite-cljs]} Issue #231.

1.2.2 - 2022-02-11

Added

• Added "defn-" to style :rod (which previously modified only "defn").

Fixed

• Fixed problem with complex :style configurations, where styles specified in a vector (in order to all be used) didn't configure properly if they also specified multiple styles if a vector. Issue #217.

1.2.1 - 2022-01-26

Added

• Added the capability to get "real" newlines (or whatever line endings are used) in strings. {:output {:real-le? true :real-le-length 20}} controls this, with "real" line endings replacing the escaped versions (e.g "\n") in strings whose length is equal or greater than :real-le-length, which defaults to 20. This should only be used when creating output as opposed to formatting code. Issue #38.

• Added a new capability to recover the adjusted range and formatted output for just that range when using {:input {:range {:start n :end m}}}. If you also include {:output {:range? true}} the return is a vector with: [{:range {:actual-start s :actual-end t}} "..."] where the "..." is the formatted output for the range of :actual-start to :actual-end in the string (file) input to zprint-file-str. This is important because zprint will always expand the given :start and :end to encompass entire top-level expressions. Using {:output {:range? true}} allows you to recover the expanded range that zprint used, as well as just the formatted lines for that range. If the :actual-start and :actual-end are both -1, that signals that no formatting was performed, the string (the second thing in the vector) is replaced by nil. This will occur when the input range included only lines for which no formatting was possible, typically blank lines or lines with only comments. The purpose of this capability is to ease integration of zprint into an extension for an editor or an IDE. You can pass in the start and end without regard to whether or not they encompass entire top level expressions, and zprint will figure out how to expand them to cover the minimally correct number of lines beyond the start and end. Using {:output {:range? true}}, zprint will tell you the lines that it figured out, and will return just the formatted output related to those lines. Note that the :actual-start and :actual-end refer to lines in the input file (string). The number of lines encompassed by the :actual-start and :actual-end will, in general, not be the same as the number of lines in the string returned as the formatted output. It will certainly be related, but may be larger, the same, or smaller. Were you to use this capability to integrate with an editor or IDE, you would replace the lines :actual-start through :actual-end in the input document with the formatted output returned in the string.

Changed

• Enhanced the processing for the experimental :rod ("Rules of Defn") style to allow similar processing except for not having blank lines between multiple arities in a function. New style: :rod-no-ma-nl, which stands for "rules-of-defn no multi-arity newline". Issue #170.

Fixed

• Fixed problem in formatting of are expressions. There were two problems, where some expressions wouldn't format with the correct number of columns, and the columns were usually too wide as well. Issue #212.

• Fixed issue where the local abs function conflicted with the new abs function added to Clojure 1.11.0-alpha4. Issue #215.

1.2.0 - 2021-11-18

Added

• {:map {:key-value-options {key-1 options-map-1 key-2 options-map-2 ...}} lets you specify an options map to be used for formatting the value of any specific key in a map. I first conceived of this as a way to solve some request in an issue from a good while ago, but I can't seem to find which issue it was, for which I aplogize to whomever triggered this capability!

• {:vector {:force-nl? true}} will cause a vector to never format on a single line, even if it would fit. Also, {:list {:force-nl? true}} will do the same for lists. There has been growing desire for something like this, and Issue #184 pushed me into actually implementing it.

• {:list {:wrap? true}} will cause a list to wrap just like vectors do by default. You could get this for a function using the :wrap function type, but now you can get it for lists in general as well.

• New styles: :require-macros-justify and :import-justify, as well as :ns-justify that includes both of them together with the existing :require-justify. Each of the -justify styles has an associated style: :rj-var, :rjm-var, and :ij-var which allows you to change the :max-variance for justification separately for each by redefining them as changes to the :style-map. Issue #166.

• New style: :quote-wrap, to be used where you want quoted lists to wrap (like vectors) and not trail down the page as lists will do by default. Issue #175.

Changed

• Problem with loss of spaces preceding inline comments at the top level. This turns out to be a bigger problem of multiple things on the same line at the top level. This fix for this entire problem now causes all of the expressions encountered by zprint-file-str at the top level to begin on a new line, as they have always done when using parse-string-all in the library. This is a behavior change for zprint-file-str (which is the basis of the standalong binaries), but of a behavior that is broken yet which has never been reported. If this change in behavior is a problem for you, please create an issue and we will figure something out. I could make this new behavior configurable if necessary, but didn't want to do that unless required, as there are already plenty of configuration options. Issue #191.

• Now using a very new version of the parser rewrite-clj for both Clojure as well as Clojurescript. The integration testing went very well. That said, be alert for subtle changes resulting from this major change.

• Made style signature1 part of the standard definition for defprotocol. This now prints different arities on separate lines. If you like it more compact (i.e., the way it was before this change), then define this for defprotocol: {:fn-map {"defprotocol" [:none-body {:style :defprotocolguide}]}}. Issue #181.

Fixed

• Expressions containing an empty list () are missing when using {:style :indent-only}. This a serious problem, and is now fixed. Issue #209.

• Documentation error in getting zprint for Linux. Testing the result with -e is wrong, now you test it with --explain-all. Part of Issue #101.

• Fixed problem where user defined fn was declared :arg2-extend, and then zprint threw an exception when formatting data that was not "extend like". Issue #200.

• Fixed :quote to change indent to 1. Issue #175.

• ##NaN, etc, now work. Used new parser! Many thanks to @lread for stepping up to support rewrite-clj and merging rewrite-cljs with it! Issue #153.

• Fixed a failure when using :range, if one of lines specified fell on an unevaluated expression, things didn't work. Issue #190.

• Namespaces maps don't work in rewrite-clj v0, but now they do since we moved to v1. Issue #199.

• zprint -e was incorrectly documented as a way to test an installation. Issue #201.

• Didn't want to print a vector on one line, even though it fit. Added :vector {:force-nl? true} to prevent this. Issue #184.

• Fixed problem with :style :hiccup affecting code formatting inside a vector. Issue #204.

• Fixed a problem where :style :keyword-respect-nl threw an exception when encountering a vector like: [#_a 1], because the first thing in the vector wasn't something that could be passed to option-fn-first. Issue #206.

1.1.2 - 2021-5-2

Added

• There was a request to justify the column after the namespaces in a :require clause of the ns. This can make any file which requires a lot of namespaces look pretty nice: :style :require-justify. Presently this is experimental, but it won't be going away, it just might change slightly as more people use it and it improves. The :max-variance it uses is 20 and comes from :style :rj-var. It can be changed thus: {:style :require-justify :style-map {:rj-var {:pair {:justify {:max-variance n}}}}} to use n as a max-variance for just the :require-justify. You can also use :style :require-pair, which is similar but doesn't include justification. It is also implemented entirely differently, so if something odd happens with :style :require-justify then :style :require-pair is a likely fallback. Issue #166.

• Maps with an odd number of values (i.e., unbalanced pairs) can cause problems with some advanced features of zprint. Now, by default, maps where the odd value is ... will work, and any map with a key equal to ... will not be sorted. This enables zprint to format maps which have been output with a ... to represent elements not present. Two new sets have been added to support this behavior. {:parse {:ignore-if-parse-fails #{"..."}}} is a set in which you can put things that show up as single elements in a map, and which will be removed for internal processing but will still show up in the output. In addition, {:map {:key-no-sort #{"..."}}} is a set of strings that are matched against keys in a map, and if any of them match the string representation of any keys in a map, that map will not be sorted. Issue #188.

• An experimental style to format defn forms differently. {:style :rod} is a first cut at someone's particular "rules of defn". The output is likely to change a bit, but if you try it and prefer this to the normal defn output, please let me know (and include what you like about it). Issue #170.

Changed

• Changed how quoted lists are handled. Previously a quoted list was handled like a regular list, where it was assumed to be code, and formatted with elements appearing the function position that appear in the :fn-map using the supplied fn-type. For elements not appearing in the :fn-map, the second argument was placed on the same line as the first, and subsequent arguments were placed beneath the second if possible. Now, because of the default configuration of :quote in the :fn-map, quoted lists are not considred "in-code", and the :fn-map is ignored when formatting a quoted list. In addition, the elements in the list are flowed with an :indent of 1. Any maps in the list will have their keys sorted, since by default maps allow sorting of keys when not in-code. If you would like quoted lists to wrap (as vectors do by default), you can use :style :quote-wrap. If you prefer to not have quoted lists handled any differently then they are today, you can configure {:fn-map {:quote :none}} in an options map, and receive the current default behavior. Issue #175.

• zprint now recognizes that any keywords in the "local" position (i.e., left hand side) of binding vector pairs are likely part of the "better cond" macro. They are now processed together, which only really makes a difference if you are trying to force the value expressions onto separate lines from the locals, in that it will not force the right-hand-sides of these better-cond expressions to the next lines. Issue #178.

• Justification has changed considerably, resulting in minor but significant changes to the default behavior. The new approach to justification looks at the variance of the lengths of the left-hand-sides of whatever pairs are attempting to be justified, and will only justify them if the variance is below the configured :max-variance, which is separately configurable for maps, binding vectors, and pairs. In addition, it will leave out up to two different lengths of left-hand-sides and justify the remaining rows of pairs if that will bring the variance below the maximum allowed. You can set the variance for :map, :binding, and :pair by setting {:binding {:justify {:max-variance 20}}} or any other number besides 20 that you want. This works as above :map and :pair as well as binding. If you want the previous justification approach, you can use {:style :justified-original} which sets the :max-variance to 1000 for :binding, :map, and :pair, yielding the previous behavior. You might find that justification is a lot more visually pleasing now with this more nuanced approach. Issue #179.

• The comment function now will not hang its arguments. If you wish to return to the previous approach, use {:fn-map {"comment" :none}}. Issue #182.

• By default any map with a key "..." will not be sorted as part of changes for Issue #188. If you wish a return to the previous behavior, include this in your options map: {:remove {:map {:key-no-sort #{"..."}}}}.

• It happens that reset! wasn't in the :fn-map. Since it is like swap!, which is :arg2, made reset! :arg2 as well.

• The are function previously didn't format any differently than any other function, which made it particularly unreadable. Now the are function is formatted as requested in Issue #81.

Fixed

• :interpose was not successfully configured from ~/.zprintrc or local .zprintrc files when using pre-built binaries or the uberjar. Issue #186.

• You can now have :next-inner inside of a :next-inner. Issue #185.

• defprotocol and defrecord are now "body" functions, which means that the indentation of the forms does not change from 2 to 1 when using :style :community. Issue #176.

• defprotocol formatting has been extended to properly handle the changes in Clojure 1.10, where :extend-via-metadata true is allowed after the component name. Issue #181.

• Loss of comments inside meta-data expressions. Issue #187.

1.1.1 - 2021-1-20

Added

Changed

Fixed

• In a very obscure case, requiring use of {:list {:respect-nl? true}}, and a hanging final parenthesis, a very short, unnamed anonymous function would disappear. This is clearly a very serious issue despite the number of specific conditions required to provoke it. Issue #173.

1.1.0 - 2021-1-16

Added

• Documentation for brew install --cask zprint. Pull request #161.

• Several new, simpler, and clearer examples in the community documentation. Issue #165.

• New options :nl-separator-all? for :binding, :map, and :pair. These are similar to :nl-separator?, but where :nl-separator? will place a blank line after a pair only when the right element of the pair was placed on the line below the left element, :nl-separator-all? will always place a blank line after every pair, regardless of where the right element of the pair ended up. There are also new styles :binding-nl-all, :map-nl-all, and :pair-nl-all which can be used to configure this approach for the respective elements (and set :indent 0 so it looks like you probably expect). :Issue #165.

• You can now define a style by changing the :style-map and invoke (i.e. utilize) the style in the same .zprintrc or the same call to set-options!.

• Several new switches and variants of the existing -w and --write switches on the distributed pre-built binaries and the uberjar. You can now "check" files (without changing them) with the -c and --check switches. Both the -c and -w switches can take modifiers, shown here modifying the -w switch (and they work with -c as well):

  • -lw like -w, but indicate which files processed
  • -fw like -w, but indicate which files changed
  • -sw like -w, but include a summary of the number of files processed and how many required a format change, as well as any errors

You can combine these, as in -lfsw as well. See here for full details, or type zprint -h for help.

Changed

• Type hint in redef.cljc to allow library build in graalVM 20.2.0. Pull request #171.

• The -e and --explain switches on the pre-built binaries and the uberjar will now only report the vales in the options map that differ from the defaults. If you wish to see the entire options map, you can use --explain-all.

• The distributed, pre-built binaries now will not replace an existing file with an unchanged file when using the -w (--write) switch, or any of its variants (see above). Previously they re-wrote the input file with the formatted output regardless of whether or not it changed. Now they do not write unchanged data back into the file.

• In Issue #124 we added the ability to have inline functions defined in some .zprintrc files, but not all of them. In addition, the graalVM binarie didn't support function definitions at all in any .zprintrc files. Now you can define functions in any .zprintrc files, regardless of the binary you are using. This is not a security issue because the inline function definitions are defined and executed with the "Small Clojure Interpreter", sci. In addition, these .zprintrc files will work in the pre-built graalVM binaries.

• The :flow function type (in the :fn-map) now supports constant pairing.

• The :moustache style has been split into two part and changed to include justification for the pairs.

• Changed the :fn-map for defexpect to use :style :respect-nl only at the top level of the defexpect.

Fixed

• Error found but not reported in .zprintrc and .zprint.edn files found with :search-config? true. Issue #167.

• Fixed two subtle hang flow problems, Issue #163.

• Cleaned up :fn-map entries for -> and ->> to not allow constant pairing at the top level of these functions, since they may well have keywords listed as functions. It is still enabled down a level inside either function. Issue #164.

1.0.2 - 2020-11-10

Added

• A new swtich for the uberjar and pre-built binaries: -w and --write. This performs in-place formatting, reading the file, formatting it, and re-writing it in place. It will work for one file or a set of file. For instance: $ zprint -w *.clj will format all of the .clj files in the current directory. If any of the files have errors, the contents of that file are left unchanged, an error is written to stderr, and the exit-status is set to 1. This runs faster (possibly much faster) than lein zprint, and does essentially the same thing. Issue #159.

• A new style: :dark-color-map, which sets both the :color-map and the :uneval {:color-map ...} to new colors which may not be your favorites, but are at least visible with a dark terminal theme.

• You can now change the things considered "constant" for constant-pairing. If you configure a one argument function as the value of :constant-pair-fn, it will be called with something to determine if it is a "constant", and therefore should be paired up. In this way, you can alter the pairing behavior of constant-pairing. Issue #56.

• A new style, :moustache, which adds vectors to the elements to be paired by constant-pairing for the function app. Issue #56.

Changed

• Once again, changed project.clj dependency and profile organization to try to keep the generated pom.xml something that cljdoc can digest successfully.

Fixed

• If a file doesn't format correctly (perhaps because of a mal-formed ;!zprint directive), the file is unchanged, the error is sent to stderr, and the exit-status is 1. Issue #160.

• All standard line endings (e.g., \n \r\n and \r) are now fully supported in all aspects of zprint. The line ending is determined by what zprint sees in the first 2000 characters of the string or file. Issue #150.

1.0.1 - 2020-9-22

Added

• Integrated new color map capabilities developed by @RingMan.
  Now, many more colors are available (including background colors), and you can now combine ANSI color values for additional flexibility. Additionally, you can now set the colors for: :char, :comma, false, :regex, :symbol, and :true. Issue #140, #152..

• Default function formatting -- you can now place a key in the :fn-map of :default, and the function-type (and possible options map) will be used for any function which does not have an explicit function-type set in the :fn-map or whose function-type in the :fn-map is :none.

Changed

• Changed from classic expectations testing framework, to development version of expectations.clojure.test after porting it to Clojurescript.

• {:map {:lift-ns? false}} -- the default for :map :lift-ns? was changed to false from true, largely because formatting deps.edn files looks ugly when the :mvn is lifted out. An alternative would have been to only lift a namespace out of a map if there were more than one key. Please submit an issue if you feel strongly about this one way or the other and we can discuss it.

• Changed default :max-depth and :max-length to be 1000000 instead of 1000, as the defexpect in zprint_test.cljc had more than 1000 expect entries in it!

• Changed :fm-map for defexpect to have :style :respect-bl so that blank lines between expect clauses are preserved.

Fixed

• Fixed problem with {:input {:range {:start ... :end ...}}} and how it interacts with {:parse {:interpose ...}}. The :interpose affects blank lines, and the range specification gathered unspecified and inconsistent blank lines at the start and end of the range. Now the range will have no blank lines at the start or end of a range (unless it is the start or end of the file), making its integration with :interpose much clearer. Issue #154.

• When formatting structures (not code), (nil nil) would output only (). Fixed. Issue #150.

• Fixed comments .zprintrc files. Issue #151.

• Fixed several small bugs in Clojurescript formatting now that more complete Clojurescript tests are easily available:

  • inline comments sometimes don't line up
  • indent-only commas missing
  • namespaced maps formatting differently in cljs from clj

• Fixed problem with negative start range, depending on end of range could throw exception. Also, now if start and end of range are in the same gap between expressions, nothing is formatted. Related to Issue #154 above.

• Fixed problem where {:map {:unlift-ns? true}} was properly distributing the namespace among the various keys, but wasn't removing the colons from keywords in the process, so that {:map {:lift-ns? true}} and {:map {:unlift-ns? true}} were not opposites. Now they are. Issue #156.

1.0.0 - 2020-6-8

Added

• Support for babashka scripts (or anything with a shebang as the first line). They format correctly, and now you can have special configuration for scripts using {:script {:more-options ...}}.

• Formatting supported for ranges of input lines. {:input {:range {:start ... :end ...}}} to specify the lines. Supported for zprint-file-str, uberjar, and precompiled binaries. Issue #122.

• :style :fast-hang, which will offer massive formatting speed-up for deeply nested code and structures, at the cost of output that takes a few more lines than classic zprint formatting and, rarely, looks a bit awkward.

Changed

• Moved to 1.0.0 per Alex's suggestion.

Fixed

• Should comments have special treatment with {:parse {:interpose "\n\n"}}. Yes, comments with no blank lines after them will never get any additional lines from {:parse {:interpose ...}}. Issue #101.

• Don't output spaces on empty lines. Now we don't output spaces at the end of any lines. Issue #130.

• Long comment lines are split when using :indent-only. Now they aren't. Issue #131.

• :respect-bl still deletes blank lines from dependencies. Actually, :respect-bl has problems with vectors. Now it doesn't. Issue #132.

• Aligned inline comments don't work right with {:style :respect-nl}. Now they do. Issue #135.

• Constant pairing count is fooled by comments, newlines. Now it isn't. Issue #136.

• Last pair in a map has comment when followed by a comment. Now it doesn't. Issue #137.

• Newline ignored when after last map element and :respect-nl used. Fixed. Issue #139.

• :respect-bl doesn't work with comments in sets. Fixed. Issue #139.

• Comments missing when in empty list. Fixed. Issue #141.

• Inconsistent indent for hanging right close paren/bracket/brace. If you have a comment as the last thing in a collection, the closing element (paren, brace, bracket) goes by itself on the next line. The question is -- where should it go on the next line. After a lot of experimentation, decided to put it underneath the opening element. See Issue #143 for some examples of why.

• zprint-file-str uberjar, and binaries all drop all but one trailing newline. Now all newlines at the end are saved. Issue #144.

• Reader conditionals don't work right with :indent-only and :respect-nl. Now they do. Issue #145.

• {:parse {:left-space :keep}} doesn't work for comments. Fixed. Issue #145.

• When comment is the last thing in a collection, :indent-only adds blank line before closing. Fixed. Issue #149.

0.5.4 - 2020-3-21

Added

• Significant new capability -- Respect Blank Lines. In the same way that :respect-nl? has been implemented for lists, vectors, maps and sets, now :respect-bl? has been implemented for them all as well. This means that whenever a blank line appears in the source, it will be "respected", and will appear in the output. However, all other formatting will be applied around any blank lines that may appear. Issue #125.

While :respect-nl? was something that you might want to configure for formatting a single function, :respect-bl? is something that is perfectly reasonable to configure for processing whole files, or perhaps all of the time in your ~/.zprintrc file. If you do that, everything will operate as normal with zprint, but if you put blank lines inside a function definition, those blank lines will continue to appear in the output. And all of the information will all be formatted correctly around those blank lines.

There is a new style: {:style :respect-bl} to simplify using this capability.

• Configure the uberjar or pre-built graalVM binaries from a URL. --url will add an options map found at the specified URL to other configuration found from the environment (for instance, ~/.zprintrc). --url-only will configure zprint only from an options map found at the specified URL. Thanks to Colin Taylor (coltnz) for the idea and the pull request for this feature. I learned a lot about Java manipulation of URLs and other http related data from his work. Issues #117 and #112.

• The built in pretty printer for Clojure, clojure.pprint, will backtranslate (quote a) to 'a, (var a) to #'a, (clojure.core/deref a) to @a and (clojure.core/unquote a) to ~a. clojure.pprint only does this when printing data structures, (which may be code), since that is all it operates on. Zprint has been enhanced to optionally perform this backtranslation when formatting structures. Use {:style :backtranslate} to get this behavior, and see the definition of that style to see how it was implemented. Even if you use {:style :backtranslate}, there is no change to the way that source is formatted since zprint has been enhanced to distinguish between source and structures in some situations. You could configure zprint to do this backtranslation for source if you wished to, though there is not a pre-defined style to enable that operation. See the implementation for {:style :backtranslate} for a hint for how to do this (or open an issue and ask). Issue #121.

• In some cases, when specifying option maps for remote invocations of zprint, boolean true and false in the options map turns into something other than true and false when zprint is invoked remotely. For instance, true can become 1 in the remove invocation, and false can become 0. This makes it impossible to invoke zprint with a reasaonable options map, since both 1 and 0 fail to validate as they are not boolean. Relaxing the validation rules would not help, as 0 is never going to be false for Clojure.

In this (rather obscure) case you could set :coerce-to-false to the value that you want to be false. If you do this the options map you specify will be searched for all values which must be boolean. If they are already boolean (i.e., already true or false), they are not changed. If they are not boolean, then if they equal the value of :coerce-to-false, they will be set to false, and otherwise they will be set to true.

In the example above, {:coerce-to-false 0} would correctly set the various boolean values. Issue #111.

Changed

• Two options map keys were previously ignored if specified in an options map on the command line: :cwd-zprintrc? and :search-config?. These keys are now examined and used if specified an options map on the command line or in a call to set-options!. Issue #120.

• The clojure.core function assert-args is now a :pair-fn.

Fixed

• Fixed an issue when .zprintrc file failed to validate when set-options! was first called, it would be cause zprint to not be initialized but no errors would be output. Now errors are output. Issue #127

• :parallel? was always set to true when running at the REPL even if it was explicitly configured false in a .zprintrc file. Now the value is not changed if there was any explicit configuration for :parallel?. Issue #128.

• While functions have been valid configuration inside of options maps for some time, they were only valid when configured with set-options!, as they would not read in properly from .zprintrc files or the command line. Now fns can be read in successfully from ~/.zprintrc and from options maps specified on the command line when using the uberjar or lein-zprint, but not when using the pre-built graalVM binaries that are distributed in the GitHub release. Functions are never allowed when reading .zprintrc files not in the $HOME directory and in options maps from URLs. In these files they are considered errors (in order to allow opening up that path with explicit configuration in the future). Issue #124.

• Fixed issue where vectors output using :respect-nl? didn't properly handle two newlines not separated by other whitespace. Issue #129.

0.5.3 - 2019-11-9

Added

Changed

Fixed

• If the style {:style :respect-nl} were used, it would not respect newlines in a binding vector (e.g., the vector for let). This should have been controlled by {:vector {:respect-nl? true}}. Issue #118.

0.5.2 - 2019-11-6

Fixed

• Bug in the graalVM binaries: zprintm-0.5.1 -e would throw an exception. Issue #116.

0.5.1 - 2019-10-29

Added

• Now you can force colored output from the zprint-filter (uberjar and graalVM binaries) by adding {:color? true} to the options map when using them. The resulting output will have ANSI escapse sequences embedded in the output, which will result in "colorized" output when displayed on a "terminal" which supports ANSI escape sequences. The output will be useless when stored in a file (unless the point is to save it for later display on a "terminal"). As part of this implementation the :color? key will now control color on all of th zprint, zprint-*, czprint and czprint-* functions. The functions beginning with "c" simply specify {:color? true} for you.

  DO NOT place {:color? true} in your $HOME/.zprintrc file, or all of the zprint-filter output (uberjar and graalVM binaries) will produce files with embedded ANSI escape sequences!

Fixed

• Removed internal use of {:style :spec} which was itself removed in 0.5.0. Issue #115.

• (zprint-fn-str) and (czprint-fn-str) didn't add specs to docstring.

0.5.0 - 2019-10-19

Because of the considerable changes in this version, moved from 0.4.x to 0.5.x.

Added

• Respect New Lines. :respect-nl? has been implemented for lists, maps, sets, and vectors. Also as a style for all of them: :style :respect-nl. This will ensure that all newlines in the input also appear in the formatted output. Additional lines may also be added to format more correctly or because lines are too long. Can be used both as a style for everything, or more frequently as a style to be used when formatting a particular function where one of the existing formatting styles isn't quite right. If you have a function "func" which would be better handled with :style :respect-nl, then an options map of {:fn-map {"func" [:none {:style :respect-nl}]}} will do that for you. Issue #75, and partially #71.

• Indent Only. Will not remove or add newlines -- will only regularize white space while preserving the content of each line. This is very different from classic zprint, and even from :style :respect-nl, above. It is available individually for lists, maps, sets, and vectors. It is also available as style: :style :indent-only, which is the recommended way to use it (if you want this capability). See the documentation for details. It will clean up the indenting and white space, while doing little else. You can do this for a whole file by specifying the options map {:style :indent-only}, or for a single function (say "func") by giving this options map: {:fn-map {"func" [:none {:style :indent-only}]}}.

• Default mode. Some people wanted a zprint that was not configurable. If you give the uberjar (or graalVM binaries) "-d" or "--default" on the command line, it will run zprint with no external configuration.
  The default configuration is what you get, and no other options are read from anywhere. In particular, the $HOME/.zprintrc file is not examined, nor is any other external configuration file. No matter who runs it, running zprint with "-s" will produce the same output from the same input file.

• -v and --version switches. Issue #94.

• -h, --help switches.

• -e, --explain switches to output configuration and show where any non-default values came from.

• :inline-align-style to align inline comments in one of three ways: :none, :aligned, :consecutive. The default is :aligned, where if inline comments are aligned on input, it will align them on output as long as they aren't separated by more than 5 non-comment lines. :consecutive aligns inline comments that appear on consecutive lines. :none doesn't align inline comments beyond what naturally happens. Issue #37.

• :next-inner in the options map of a :fn-map vector. This allows you to set some configuration for only one level of processing. See the readme section "Altering the formatting inside of certain functions" for details. Implemented to resolve Issues #71 and #86.

• :option-fn and :fn-format in :vector. Thanks to some creative thinking and clever implementation by @milankinen (Matti Lankinen), the formatting available when formatting functions in lists is now also available for formatting vectors. He implemented this to allow formatting hiccup vectors in a standard and readable way, but there are plenty of other opportunities to format vectors in useful and clear ways. A huge thanks to Matti for contributing this to zprint.

Changed

• If there was an error on the command line, either in the new switches (see above), or the options map, the uberjar and graalVM binaries used to format the input and report the error after formatting. Now they do not format at all, and just report the error.

• Configuring zprint with environment variables or Java properties has been deprecated for a good while, and is now no longer supported. As part of this change, the {:additional-libaries? ...} key is now not supported (or allowed).

• Removed {:style :spec}, since it has been deprecated since the relevant spec functions were enhanced to force {:list {:constant-pair-min 2}}.

Fixed

• If you entered:

(;comment
 function this that and the other)

then `function` would not previously have been recognized if
it was defined in the `:fn-map`.  The "thing" in the first part
of the list was what was looked up, and if it was a comment,
that was too bad.  Now zprint will look up only actual symbols,
not comments (or newlines).

• If an inline comment wrapped, it would be on a line by itself. It would be aligned with the actual inline comment above it. But the next time you ran zprint on the file, the wrapped inline comment would be seen a a full line comment, and moved (probably left) to be where a full line comment should be. Fixed this, so that a comment with nothing else on the line that is aligned with an existing inline comment will stay aligned with that inline comment. Issue #67.

• In vectors, a full line comment could become an inline comment because of the way they wrapped. Fixed this. Issue #39.

• Added (require clojure.string ...) to spec.cljc. Issue #109.

• Removed nested conditional compilation in core.cljc. Issue #110.

0.4.16 - 2019-06-12

Added

Changed

• Implemented :unlift-ns?. If this is specified for a map when formatting code (the only place it makes sense), a map specified with a namespace at the front (e.g., #:x{:a :b :c :d} will be formatted with the namespace distributed among the keys, as in {:x/:a :b, :x/:c :d}. Issue #80.

Fixed

• Fixed record output to be something that can also be read back in. Replace "/" in record-type with ".". Issue #105.

• Fixed serious bug in handling of pairs with comments: if a comment was the last thing in a series of pairs, the following right parens would be missing! Issue #106.

• Fixed bug in flow indentation underneath things like "#(...)". Issue #103.

• Fixed bug where files that ended with a newline didn't after being processed with {:parse {:interpose "\n\n"}}. Issue #100.

• Fixed bug where namespaced maps in source code didn't format correctly. Issue #104.

• Fixed code to correctly detect when running in a REPL and enable parallel operation. It was always enabling parallel operation before, which will cause a main program using zprint to not exit correctly unless you call (shutdown-agents). Now it only enables parallel operation automatically if you are in a REPL, or if you call (set-options! {:parallel? true}). Issue #107.

• Removed LC= from build.zprintm, since that apparently didn't work for the macOS version of graalvm. Issue #96.

0.4.15 - 2019-01-14

Fixed

• Fixed cljs build warning.

0.4.14 - 2019-01-13

Added

• New configuration capabilities. Everywhere zprint looks for a .zprintrc file, it will also look for a .zprint.edn file if it does not find a .zprintrc file. Issue #83

• If the .zprintrc or .zprint.edn file in the $HOME directory has :search-config? set to true, zprint will look for a .zprintrc or .zprint.edn file in the current directory, and if does not find one there, in that directory's parent, and will continue to look in directories up the tree until it either finds a configuration file or runs into the root directory. If :search-confing? is true, :cwd-zprintrc is ignored.

• Major new capability to rewrite zippers during formatting to do things like sort dependencies. This is alpha/experimental, which is to say, the capability may go away or change significantly. It has had less testing than usual, but has been tested in both Clojure and Clojurescript. You can access this capabiity for Leiningen project.clj files by using {:style :sort-dependencies}. Issue #76

• Information on using Clojure CLI. Thanks for pull request #82

Changed

• Enhanced formattng of :binding functions by allowing empty body form: (let [a b c d]). This will successfully format as a :binding form. Issue #73

Fixed

• Wrapped a dec function in an fnil to avoid an overly enthusiatic Clojurescript warning in Clojurescript 1.10.439. Issue #91.

• Added build parameter to graalvm binary builds which helps in handling a broader character set. Uberjar with JVM worked anyway.

• Fixed zprint-file-str and {:parse {:interpose "\n\n"}} which interacted badly. Comments (including ;!zprint comments) would have an extra blank line. Issue #84.

0.4.13 - 2018-11-21

Changed

Fixed

• Fixed exception when formatting complex :arg2-extend functions, like defrecord, which have functions that don't start with simple symbols or keywords. Issue #72.

• Fixed exception when formatting a parsed map with keys shorter than 2 characters. Doesn't affect structure formatting, just maps that show up in strings that we parse (as in formatting source). Issue #74.

0.4.12 - 2018-11-6

Changed

• Allow any number of semicolons when checking for ;!zprint. Thus, ;;!zprint and ;;;!zprint also work.

Fixed

• Return non-zero exit status from uberjar (and graalvm binaries) if it doesn't format correctly or the options map on the command line is bad.

• Place all exception information for the uberjar (and graalvm binaries) on stderr, not stdout.

0.4.11 - 2018-10-28

Changed

• Added zprintm-0.4.11 MacOS zprint filter to release. Thanks to Martin Klepsch for pointing out a "ce" version of graalvm exists for MacOS. Issue #66.

• Verified correct operation in Clojure 1.10.0-beta3.

• Moved to rewrite-clj 0.6.1.

• Added cond-let as a :pair-fn.

• Added a bunch of better doc-strings, doc/cljdoc.edn to make doc work better with cljdoc.

• Moved lumo and planck into .cljs files so that they work with cljdoc.

Fixed

• Major bug fix, (or enhancement, depending on your viewpoint). You can now run multiple zprints in the same JVM at the same time, with the restriction that they all need to be doing the same 'kind' of zprint, either all parsing strings (typically formatting code), or all operating on Clojure(script) structures. Can't mix them. But you can use pmap to get zprint-file to operate on a bunch of files at the same time (which was the use case that prompted this change). Issue #63.

• Minor bug where #( stayed green inside of syntax-quoted structure, instead of turning red like all other parens to.

0.4.10 - 2018-7-28

Changed

• Put ^:no-doc in various places to prepare for codox. Put codox configuration in project.clj.

• Added capability to have a per-project .zprintrc file. If your $HOME/.zprintrc file has {:cwd-zprintrc? true} in its options map, then zprint will also look for a .zprintrc file in the current working directory.

• Added capability to have an options map associated with a function classification. Support for Issue #46, as well as changes to the classification of defproject, s/def, and s/fdef. The changes to s/def and s/fdef now let specs format correctly by default, and the :spec style is no longer necessary (but still there for compatibility).

• Added new style :how-to-ns, to format ns declarations as in Stewart Sierra's "How to ns". Issue #46.

• Documented how to keep zprint from configuring from a .zprintrc file, for those uses where the options should not be affected by any local variations: (set-options! {:configured? true}) must be the first use of zprint.

Fixed

• Tests now are independent of any $HOME/.zprintrc or local .zprintrc.

• Don't close stdout when zprint.main/-main exits, flush instead. Fixes use of zprint.main by clj command. Issue #57.

• Fixed appcds checking for java version, now it doesn't. Issue #54.

• Fixed extra spaces between newlines when using (:style :map-nl}. Issue #59.

• Fixed problem with #_#_ syntax, where it was missing the second element. Issue #58.

• Fixed some instructions for graalvm -- set script executable, added #!/bin/bash to build.zprintm.

0.4.9 - 2018-5-14

Changed

Fixed

• Back off to Clojure 1.8, since 1.9 and lein zprint don't agree with each other.

0.4.8 - 2018-5-13

Changed

• Both :max-length and :max-depth are now offical parts of zprint. They are not experimental anymore.

• Added {:max-length [n m]} capability -- different length boundaries for different levels. Sometimes you want to see all of the top level keys in a map, and just a little of the complex values.

• Added ability to set the character for max-depth: :max-depth-string is default to "##", but could be anything (e.g., "#").

• Added match and matchm to the :fn-map as :arg1-pair-body. These are from core.match. Issue #50.

Fixed

• Removed (:gen-class) from the .cljs version. Issue #51.

• {:max-hang-depth n} works along with {:max-hang-count m} to decide to even try to do a hang. Having these be 3 and 4, respectively, makes some maps format faster, but also messes up perfectly reasonable maps to the point that they are really just wrong. So, set :max-hang-depth to 300 as a default to disable this optimization. Issue #48.

• {:max-length n} didn't work for lists, and had a number of other problems that are now fixed.

• {:max-depth n} also had issues. In particular, 0 and 1 were the same. Now 0 will print non-collections, but all collections are ##.

• Changed :community style to leave cond-> as :arg1-pair-body, instead of moving it to :none-body. Issue #49.

• (zprint nil :explain) would output the options map in color, even though it was zprint and not czprint. Now it doesn't.

0.4.7 - 2018-2-19

Changed

• Rewrote line-lengths to use iterative instead of really complicated functional approach. Saved maybe 5% on time. Still complex, but more understandable and slightly faster.

• Added performance optimization :hang-avoid in :list which will not even bother to try a hang if lots left to do and very close to the right margin. Didn't change output of anything tested or zprint code itself, but saved up to 30% on very pessimal functions. Not so much on normal functions, but still a win for those tough ones.

• Added two new styles: :no-hang and :all-hang, to allow configuring zprint without hangs or with hangs. While I think that code formatted without hangs looks rather unlovely, some folks find the way it looks to be a worthwhile tradeoff for speedier performance without hangs, particularly in Clojurescript. So, now, you can configure that easily one way or the other.

Fixed

• Fixed bug in configuration where :parallel? wasn't being set to true even for repl use (or in lein zprint). Saved up to 20% in lein zprint, which is actually quite a bit.

0.4.6 - 2018-1-10

Changed

• Added {:vector {:respect-nl? true}} support, so that hand-formatted vectors can be preserved on ouput.

• Added {:vector {:option-fn-first #()}}, which allows definition of a function to be called with the first element of every vector, and returns an options map to be used to format that vector and all that it contains. Added primarily to allow targetted use of :respect-nl?, but is fully general and can certainly be used to enable other capabilities.

• Added new style :keyword-respect-nl, which will use the two new capabilities above to preserve existing newlines in vectors that start with a keyword (as hiccup and rum HTML data do).

• Extended zprint-file-str to accept an options map.

• Added zprint-file and zprint-file-str to the external API, and documented the use of ;!zprint directives in the zprint readme. (It was already documented in the lein-zprint readme.)

• Added tests for zprint-file-str, which previously was only tested by lein-zprint.

Fixed

• Additionally, changed zprint-file-str to leave the configured options unchanged after processing. It used to leave them around after processing to be retrieved with (get-options), possibly distorting the next call to any zprint function.

0.4.5 - 2017-12-9

Changed

• Added support for rum macros defc, defcc, and defcs, using :arg1-mixin function tag. See the readme for details of what this looks like. Issue #41.

• Add swap! as :arg2, and with-redefs-fn as :arg1-body.

0.4.4 - 2017-10-26

Changed

• Added :color? boolean to options map, which will allow you to select color (or not) regardless of whether or not you use czprint or zprint.

• Added deftest and defexpect as :arg1-body functions in the fn-map.

Fixed

• Added inline-comment support, on by default. Controlled by {:comment {:inline? true}}. They will also be wrapped. Issue #33.

• Fixed problem with focus interacting badly with comment wrapping, where a wrapped comment would change the path for later elements if it wrapped. Now wrapping is performed after all focus operations are completed.

0.4.3 - 2017-10-10

Changed

• Added support for focus on output. This allows you to input an expression parsed with rewrite-clj and point at some expression down inside of the overall expression, and get that printed with "focus" (currently a rather garish reverse-video, but that will probably become configurable in the fullness of time).

• Added support for outputing only some lines from printing an expression, and replacing them with a configurable string to indicate an elision. Documentation on this is coming but not here yet.

• Upgraded support to Clojure 1.9-beta2 and Clojurescript 1.9.946.

• Upgraded rewrite-clj to 0.6.0.

• Added :syntax-quote-paren to the :color-map and the :uneval :color-map. By default this turns the parens :red inside of syntax-quoted expressions.

• Made assoc-in and doto both be :arg1 functions.

Fixed

• Fixed a bug where an :arg2* function which wasn't able to hang-one threw an exception. Pretty rare, needs a very deeply nested function. cgrand/poucet did manage to expose this problem.

• Map with long keys and :justify? true does not become justified. It is worth noting that :justify? also has a heuristic aspect, where if the justification takes more width than is available it will fall back to unjustified. But this also exposed a bug where the handling of the width didn't work in situations were you turned off commas. This is now fixed. Issue #31.

• zprint nil prints a blank line, not nil. Issue #32.

• The latest versions of lumo and planck both have upgraded to tools.reader 1.0.5. Unfortunately, rewrite-cljs 0.4.3 depended on an internal API for tools.reader that changed from the alpha version used by rewrite-cljs 0.4.3, but we've gotten a new version, rewrite-cljs 0.4.4, which now works with tools.reader 1.0.5, so both planck 2.8.1 and lumo 1.8-beta now work again with zprint. Issue #30.

0.4.2 - 2017-6-22

Changed

• Changed printing of :pair-fn functions (i.e. cond) so that if there is more than one pair (and if there weren't, why would you use a cond), it won't print it all on one line. If you don't like this, {:remove {:fn-gt2-force-nl #{:pair-fn}}} will turn it off.

• Added ability to lift namespaces out of maps with namespaced keys. Controlled by :lift-ns? and :lift-ns-in-code? in :map. Will not lift namespaces when specified implicitly in strings. That is, will not try to figure out correct namespace for ::a, since when parsing a file this can't reliably be known. Note that #::{} maps are not currently supported, and cannot be handled when using the :parse-string? true input approach.

Fixed

• A :>> in a condp wasn't handled correctly (i.e., at all). This made the pairing up of condp clauses all wrong after that.

• A problem when using a "modifier" in an :extend (e.g., "static" in defui) -- the modifier and the thing after it were always formatted onto the same line, even if they didn't fit. Now, if they don't fit together, they don't end up on the same line.

0.4.1 - 2017-5-18

Changed

• Added capability to pass options map to zprint-filter by changing the zprint.main namespace to handle a null string, and changing appcds script to create a slightly different script.

• Moved to cljs.spec.alpha for Clojurescript too. Zprint now requires Clojurescript 1.9.542 or later.

• It is now not even accept to set :parallel? to true in Clojurescript, as the spec has been changed to consider this an error when you try it.

• Added new capability: sorting for sets. Similar to maps, now {:set {:sort? true :sort-in-code? false}} is the default. When you have sets with a lot in them, it is nice to see them sorted.

• Added new capability: key-value-color for maps. Where key-color colors a particular key that matches the color that you specify, key-value-color will also match a key, but instead of affecting the color of that key, it merges in a color-map which is the value of that key into the current color-map, affecting the printing of everything in the value. Which may be a constant like a string or a keyword, but might be an entire complex expression. This only works for maps.

• Removed {:pair {:indent 0}} from the :spec style. If you like your pairs non-indented, feel free to add it to the style yourself. Put this: {:style-map {:spec {:pair {:indent 0}}}} in your ~/.zprintrc.

Fixed

• Fixed Issue #27, where missing function errors came up when using zprint in non-self-hosted Clojurescript.

• czprint-fn will now print specs for any function that has them. It will add them to the end of the doc-string (much like doc will do, but it will also pretty print them). In particular, try let or ns or even defn. The defn ones are a little funky, as the indent isn't all that lovely.

0.4.0 - 2017-5-5

Changed

• Changed to use clojure.spec.alpha instead of clojure.spec. Now, the 0.4.0 version runs in 1.9-alpha16, but not 1.9-alpha15. It still runs in 1.8 as well, but requires a different supporing library -- see the readme.

Fixed

• Issue #26 -- now it runs in 1.9-alpha16.

0.3.3 - 2017-5-4

Changed

• Added appcds script to set up zprint-filter with less opportunity for errors.

• Added instructions for how to use zprint-filter with emacs and Sublime Text 2 or 3.

• Changed default for :extend to {:flow? true}, to fix issues in extend-protocol, reify, and extend where things were coming out on the same line when they should not.

• Created :arg2-extend to better handle deftype and defrecord.

• Changed fn-type of defprotocol to :arg1-force-nl, so that it doesn't hang several small forms.

Fixed

• Issue #25, where additional () were added when formatting a (fn ...) when the (fn ...) was in a Clojure data structure. This didn't happen when formatting source in files or with the zprint-filter.

0.3.2 - 2017-4-18

Changed

Fixed

• Issue #23, where (czprint-fn defn {:style :justified}) would not produce any output. This bug was added in 0.3.0.

• More work on dependencies to make it easier to release.

0.3.1 - 2017-4-10

Fixed

• Fixed dependencies so that [clojure-future-spec "1.9.0-alpha15"] is actually required to use zprint on Clojure 1.8, and doesn't get brought along into usage on 1.9 by default.

0.3.0 - 2017-4-9

Changed

• DEPRECATED configuration from environment variables and Java system properties. Still available by adding a library to your dependencies. File an issue if you care about this!

• DEPRECATED :auto-width capability. Still available by adding a library to your dependencies. File an issue if you care about this!

• Added zprint-filter uberjar and documentation. Uberjar is available as a "release" on Github.

• Moved to clojure.spec for all versions, requiring a single additional library when using Clojure 1.8.

• Added futures to fzprint-hang-remaining, helped a lot.

• Used pmap to handle two-up generation, helped a lot.

• Added transient to some vectors being built. Sometimes it helped, somtimes it didn't. Kept the ones where it helped.

Fixed

• Dates in this file.

0.2.16 - 2017-2-27

Changed

• Added zprint.lumo namespace to support zprint filter. Only used when initiated by lumo.

• Added zprint.planck namespace to support zprint filter. Only used when initiated by plank.

• Replaced function-by-function destructuring with with-redefs in both zutil.cljc and sutil.cljs. Added new namespace zfns to make this work. Resulted in small but significant speedup.

• Major changes to Clojurescript port -- removed Prismatic/Plumatic Schema and replaced it with spec for checking options maps. Refactored option map validation into two additional namespaces. Now zprint will work in self-hosted Clojurescript, and we are ready for a complete move to spec when we can require 1.9.

0.2.15 - 2017-1-24

Changed

• Added {:map {:key-color {:key1 :color1 :key2 :color2 ...}}} which will color the specified keys with the specified color.

• Added {:map {:key-depth-color [:level-1-color :level-2-color ...]}}, which will color the map keys based on depth, not their type. The :key-color {:key :color ...} map will override any values from the :key-depth-color map. EXPERIMENTAL feature, might go away.

0.2.14 - 2017-1-22

Changed

• Added a new capability to :extend, :modfiers #{"static"}, which will allow "static" to appear on the same line as protocol or type when formatting extends. You remove elements from the set with {:remove {:extend {:modifiers #{"static"}}}} just like the other sets are changed. Issue #10.

• Added defui to the :fn-map as :arg1-extend so that when formatting Clojurescript om/next code you get the static elements to format correctly. Issue #10.

• Added several new styles: :map-nl, :binding-nl, :pair-nl. Also added substantially to the documentation of styles in the readme.

• Added the first tests for the :explain output, at least the :value part of it.

Fixed

• A problem when adding an element to any set in the options map, where the element was added, but the :explain output was incorrect. Issue #21.

0.2.13 - 2017-1-19

Changed

• Added support for :hang? in :extend, so that you can control whether or not the function definitions end up formatting as a hang after the type. :hang? defaults to true (and this is not a change), but can now be turned off to force a flow. Note that if a type and function defnition all fit on one line, that :hang? false will not affect it. If you want them on separate lines you need to use :flow? true, which will override the normal desire to print things on one line.

• Altered the meaning of :nl-separator?. In its initial release it would place an extra new-line whenever a pair took more than one line. Now, it only places an extra new-line when the pair did an actual flow, so that a multi-line hang does not trigger an extra new-line. The previous capability would be easy to add if there is any interest in doing that.

• Added the ability to set multiple styles at once, with a sequence of style keywords instead of just a single keyword. Also added some great style tests. Go expectations!

Fixed

• The new :nl-separator? capability does not work right with lists that have constant-pairs at the end but regular s-expressions earlier. If the earlier s-expressions take more than one line, then they get blank lines below them which is wrong from several different perspectives. Now multiple line s-expressions don't ever trigger blank lines, since the :nl-separator? blank lines are triggered by a pair with the right hand part of the pair being formatted with a flow. Issue #17.

0.2.12 - 2017-1-9

Fixed

• Accepted pull request to change function style for "ns" to :arg1-body. Thanks to pesterhazy!

• Accepted pull request to change function "!=" to "not=". Thanks to mynomoto!

0.2.11 - 2017-1-8

Changed

• Removed :indent n from :pair-fn, as it wasn't used.

• Changed cond and alt to be :pair-fn, and not :pair, since :pair was supposed to not be a function style on its own, but a configuration element for functions that have pairs in them, like :arg1-pair and :pair-fn.

• Added capability to produce blank lines after pairs where the second element of the pair doesn't fit on the same line as the first. The options key for this is :nl-separator?, and is by default false for all occurances of this configuration option. It is supported for :map, :extend, :binding, and :pair formatting. This was requested for :map, but the implementation also allows configuring :extend to produce a commonly used format as well. Issue #12.

• Added :flow? options key for :map, :extend, :binding, and :pair. By default false for all. When enabled, this will cause the second element of every pair to be formatted on the line after the first element, thus acting line a "flow". See the readme for differences between :flow?, :hang?, :force-nl?, and :nl-separator?. This was largely added to work together with :nl-separator? in :extend, but also has uses elsewhere. :flow? will override the indent checks in when printing things two up, so that it will force a flow even though the first element of the pair is so short that it doesn't make any sense. Just because it is so confusing if it doesn't do that.

Fixed

• Fixed constant pairs to be formatted under control of the configuration for :pair, which now includes :flow? and :nl-separator?.

• Serious bug where whitespace after left parenthesis causes either the first or second element in a list to be repeated if the list doesn't fit entirely on one line. Issue #16.

0.2.10 - 2016-12-23

Changed

• Added two new sets in the config, :fn-gt2-force-nl and :fn-gt3-force-nl into which you can put fn-styles so that they will never print on the same line even if they could. The gt2 is if it has greater than 2 args and the gt3 is likewise if it has greater than 3 args. Issue #3.

• Refactored zprint-file, pulling out zprint-file-str so that boot-fmt can use it. Added zprint-file-str to the API. Issue #7.

• Added :arg2-fn to support proxy.

• Added :arg1-force-nl to support s/def. Put several additional spec functions into the :fn-map so that spec files now pretty nicely.

• Added experimental options to try to speed up map printing at the repl, which slows down when the maps are really deeply nested. :max-hang-span :max-hang-depth :max-hang-count These are experimental, and will probably not keep their current names or location in the options map. They are not documented. Do not depend on them.

• Added :flow and :flow-body to force some functions to never hang. Largely added for :require in ns declarations. Issue #5.

• Added :require to the :fn-map as :force-nl-body so that it will hang all of the requires. If you want it to pull them all over to the left and not even have something on the same line as the :require, define it as :flow-body. Issue #2.

Fixed

• Added a capability to remove elements from a set, now that sets in the config are more common. Also, fixed it so that you can specify additional members for a set, and they will be merged into the set, instead of replacing the entire set. Issue #4.

• Fixed :fn-force-nl to accurately match the function type, and not do it after the -body has been removed. Issue #6.

• A bug in :arg1-extend where, when used for proxy, some of the function was getting dropped and not output. Since proxy was not defined as :arg1-extend, this is not as serious as it might seem, but it was not good.

0.2.9 - 2016-11-13

Changed

• Added :force-nl and :force-nl-body along with :fn-force-nl set in configuration to force some functions to never format onto the same line. Primarily for -> and ->> and other similar functions where most people want to see them listed vertically, not on the same line.

• Made keywords format as functions, since they are. This cleans up ns declarations a bit at the start of files.

Fixed

0.2.8 - 2016-11-9

Changed

• Added documentation on experimental features: auto-width, max-length, max-depth.

Fixed

• Fixed bug in handling of :left-space :keep and :drop which involved a complete rework of options map handling in process-multiple-forms and zprint-str-internal and czprint-str-internal. Should have fixed a number of other not yet encountered bugs.
• Fixed issue in fzprint-hang-remaining where if the hang-expand value was exceeded it wouldn't bother to hang. Which interacted badly with the hang-if-equal-flow? capability. Added test to ensure this stays fixed.
• Fixed bug where if the first element of a pair was a map that got a comma, and the second element of the pair ended up with a hang, the first line of the hang would have the wrong indent because the size of the first element of the pair was calculated incorrectly.
• Fixed a bug where :map {:comma? false} didn't actually work if the map fit on one line. Worked fine for multi-line maps, not one line ones.

0.2.7 - 2016-11-1

Changed

• Changed hang-expand to 4.0 most everywhere
• Changed fn-map of -> to :none-body, and catch to :arg2
• Considerable changes to support Clojurescript. All files now .cljc
• Added {:file {:left-space :keep}} or :drop to handle spaces on the left of function definitions.
• {:parse-string? true} will always trim spaces off of what is being parsed.
• (:parse-string-all? true} will parse multiple forms in one string. This does not support ;!zprint {} comments.

Fixed

• Fixed bug where -> affected :arg1 status of more than immediate args
• Fixed issue #1 with incorrect indent on 3rd arg to (fn ...)

0.2.6 - 2016-10-10

Changed

• Changes to support :reader-cond {:force-nl? boolean}
• Added defrecord to function categorization.

0.2.5 - 2016-10-06

Changed

• Changed heuristics so that if hang and flow take the same number of lines, then it will hang. Can be adjusted with tuning.

0.2.4 - 2016-10-05

Changed

• Reworked README.md, internal TOC, added information about several existing capabilities. Alphabetized the configurable elements.

0.2.3 - 2016-10-01

Changed

• Removed :list {:indent-body n}, made it :indent. Made :indent-arg optional, uses :indent if it doesn't appear.
• Added :support keyword (replaced undocumented :explain-all).
• Added documentation for :justify? in readme.
• Added :key-ignore and :key-ignore-silent for :map formatting.
• Added :to-string? for :record formatting.

Removed

• Removed (get-options) from zprint.core, took need for it out of readme.

Fixed

• A problem where the last newline in a file was getting lost.
• A problem where a comment inside of a fn starting in col 0 was getting an extra newline.

0.2.2 - 2016-09-24

Changed

• Straighten out core and config namespaces for API
• Added :old? and :format to options map for lein-zprint

0.2.1 - 2016-09-22

Changed

• Moved some configuration functions into zprint.core to match readme.

• Put ALPHA banner in readme.

• Fixed a bug where it wouldn't configure if no ~/.zprintrc file.

0.2.0 - 2016-09-21

Added

• Initial project commit.