chore: Update vendored sources to igraph/igraph@03760a09cc7195dfaad3a1e60cff1632aa6a6118 #1470

szhorvat · 2024-08-22T09:18:08Z

@krlmlr For revdepcheck purposes, as agreed. Somehow I don't see any deprecation warnings, despite igraph_minimum_spanning_tree_* functions being deprecated. @Antonov548, are deprecation warning disabled now? I don't see where.

aviator-app · 2024-08-22T09:18:11Z

Current Aviator status

Aviator will automatically update this comment as the status of the PR changes.
Comment /aviator refresh to force Aviator to re-examine your PR (or learn about other /aviator commands).

This PR was merged using Aviator.

See the real-time status of this PR on the Aviator webapp.

Use the Aviator Chrome Extension to see the status of your PR within GitHub.

szhorvat · 2024-08-22T09:19:48Z

Why are checks not running? Do I need to mark as ready-for-review? EDIT: No, checks are still not running.

R/aaa-auto.R

szhorvat · 2024-08-22T09:23:34Z

R/aaa-auto.R

@@ -1370,6 +1370,27 @@ feedback_arc_set_impl <- function(graph, weights=NULL, algo=c("approx_eades", "e
  res
 }

+feedback_vertex_set_impl <- function(graph, weights=NULL, algo=EXACT_IP) {


@Antonov548 Again, algo=EXACT_IP is invalid code, which needs to be dealt with, either by setting up translation for the new type, or removing code generation for this function temporarily. Exposing this function should be really easy, as it's near-identical to the feedback arc set one (so it's a copy and paste job). Ref: #1447 I recommend exposing it right away.

szhorvat · 2024-08-22T09:33:38Z

Disabled codegen for all distances functions, since these are exposed manually, and moving them to autogen is unlikely to happen in the near future due to the custom interface. The extra commit finally allows checks to run.

Antonov548 · 2024-08-22T11:19:26Z

@Antonov548, are deprecation warning disabled now? I don't see where.

@szhorvat deprecation warnings are disabled in igraph_config.h in order to pass CI and CRAN. You can define IGRAPH_DEPRECATED_ENUMVAL to see deprecation warnings.

szhorvat · 2024-08-22T11:22:39Z

@szhorvat deprecation warnings are disabled in igraph_config.h in order to pass CI and CRAN. You can define IGRAPH_DEPRECATED_ENUMVAL to see deprecation warnings.

I think this is perfect for now.

krlmlr · 2024-08-22T20:06:18Z

Checks run in #1472. I have no idea what's going on.

krlmlr · 2024-08-22T20:06:41Z

No, they are not!

szhorvat · 2024-08-24T23:45:18Z

Tests pass, if revdepchecks are fine, this can be merged @krlmlr

krlmlr · 2024-08-29T19:36:36Z

I currently can't run revdepchecks...

szhorvat · 2024-08-31T12:45:49Z

I currently can't run revdepchecks...

That's alright. Can we merge this though, even without doing the revdepchecks within the PR? We'd need to deal with revdepcheck breakages either way. Merging this now opens the door for exposing new functionality.

krlmlr · 2024-09-01T03:28:24Z

What functionality would you like to expose? Can we do it inside this PR? As long as it's only new features, it shouldn't make a difference for the revdepchecks.

It's easier from an organizational perspective to delay the merge until we can run the revdepchecks.

szhorvat · 2024-09-01T07:05:01Z

Can we do it inside this PR?

Yes.

szhorvat · 2024-09-05T09:00:58Z

I'm going to have a look at these now.

szhorvat · 2024-09-05T09:04:18Z

I want to refactor R_check_int_scalar() and similar functions to be macros instead. This will allow error messages to point to the exact location of the failure in the sources. Any objection, @Antonov548 @krlmlr ?

krlmlr · 2024-09-05T09:08:06Z

Good idea re macros, no objections from me. What's the impact on the size of the compiled library?

szhorvat · 2024-09-05T09:18:35Z

Convenience link to revdepchecks README:

https://github.com/igraph/rigraph/blob/e04c35515df8ebfdfb7be49f94ee85d3f00d8b48/revdep/README.md

manynet

They pass a vector for the islands.size parameter of sample_islands(). This is invalid: it must be a scalar integer, and this has always been so.

I don't recall any recent changes in igraph that would cause this failure. We did add more stringent checks in 2.0.0, but that should have caught this misuse much earlier.

The fix needs to be done in the other package. EDIT: Notified, stocnet/manynet#86

MetaNet

???

Doesn't seem related to C core changes, I won't dig deeper at this time.

Should probably be investigated further.

netropy

They are passing NaN (or NA) values as edge weights when finding shortest path lengths. This has always been invalid.

Again, as I recall, the check for this was added a long time ago, not recently, so I am surprised that the failure appears only now.

EDIT: There seem to be no call to igraph::distances() in netropy. Hypothesis: the distance calculation might be done indirectly, through ggraph, potentially because it uses Kamada-Kawai, which then internally calls igraph_distances_dijksta(). I have not verified this. The change that triggers this failure may have been in a different package than netropy. All that said, using NA/NaN edge weights will be problematic/invalid with almost all igraph functions.

The fix needs to be done in the other package.

priorCON

Might need more diggint, but overall it looks like they expect a specific result for some sort of community detection (didn't check which one).

I did make a fix to igraph_community_multilevel() (cluster_louvain() in R) which causes it to return a different result for the same random seed. We do not guarantee the same result for the same seed across bugfix versions—this would prevent many kinds of bugfixes. We only guarantee the same statistical properties.

The fix needs to be done in the other package. EDIT: Notified, cadam00/priorCON#1

R6causal

???

Found the following significant warnings:
  Note: possible error in '.to(l[i])': unused argument (l[i]) 
  Note: possible error in '.from(l[i])': unused argument (l[i])

Does this ring a bell @maelle ? You were modifying these functions, as I recall.

simcausal

???

Doesn't look igraph related.

skynet

Looks like this is due to a deliberate deprecation, fix needs to be done in other package. CC @maelle

szhorvat · 2024-09-05T09:19:47Z

What's the impact on the size of the compiled library?

Probably very small, but needs to be measured, which should be easy to do. The checks themselves should add very little code. The large part is the error string, but the compiler consolidates duplicate strings.

maelle · 2024-09-05T12:11:16Z

Reg skynet ropensci/skynet#11, will also email a reminder

krlmlr · 2024-09-05T12:59:29Z

Only priorCON seems to be a new problem in this branch: f-revdep-2...try/c-core-2024-08-22

krlmlr · 2024-09-05T12:59:51Z

The other packages are also a problem in the main branch.

schochastics · 2024-09-06T04:13:46Z

@szhorvat we are on it and will report back

schochastics · 2024-09-06T04:31:19Z

Turns out there was a call to graph_from_adjacency where the input matrix contained NA weights on the lower.tri to indicate symmetry. This is being fixed in a pull request (termehs/netropy#4)

szhorvat · 2024-09-06T07:22:15Z

Thanks so much @schochastics and @termehs!

termehs · 2024-09-06T15:13:40Z

Thanks all! New netropy version on its way to CRAN 😊

szhorvat · 2024-09-12T08:59:08Z

@krlmlr I had to make a fairly crude change to the C core to deal with potentially incorrect results (see igraph/igraph#2674). I would like to request another round of revdepchecks for this PR. I'll talk to packages that are affected by this change. Hopefully there are not many.

szhorvat · 2024-09-12T09:00:57Z

We still have the weird issue that CI tests don't get run for C core updates. I did run R/igraph tests locally after this change and they pass.

szhorvat · 2024-09-17T09:26:43Z

Can we merge this please (and have another revdepcheck round)?

krlmlr · 2024-09-18T05:03:59Z

Running checks now.

aviator-app · 2024-09-18T18:57:03Z

This pull request failed to merge: PR cannot be automatically rebased, please rebase manually to continue. Once the issues are resolved, remove the blocked label and re-queue the pull request. Note that the pull request will be automatically re-queued if it has the mergequeue label.

Additional debug info: Failed to rebase this PR onto the latest changes from the base branch. You will probably need to rebase this PR manually and resolve conflicts).

krlmlr · 2024-09-18T18:57:06Z

No regressions, good to go. Thanks!

…hese are exposed manually, and this is unlikely to change for a while

feat: allow NULL for cut_prob parameter of motif finding functions closes igraph/igraph#2555 docs: fix inaccuracy/typo in feedback_vertex_set() docs [skip ci] fuzzer: update to libxml2 2.13.3 [skip ci] chore: spelling in changelog [skip ci] ci: temporarily disable ppc64 and s390x builds on Travis as the Travis infrastructure is currently broken chore: bump cmake policy max fix: fix type of result argument in igraph_feedback_vertex_set() chore: update changelog feat: igraph_feedback_vertex_set() refactor: refer to igraph in a uniform way in headers, and minor code cleanup doc: fix typo [skip ci] chore: clarifying comment for cycle finding [skip ci] refactor: work around GCC warning in feedback arc set implementation chore: add missing copyright header to new feedback arc set test ci: disable emulated checks on push; they can be started manually instead ci: test on multiple architectures using emulation refactor: further minor cleanup in feedback arc set refactor: clean up Eades algorithm implementation for feedback arc set refactor: rename IGRAPH_FAS_EXACT_IP_TO to IGRAPH_FAS_EXACT_IP_TI and expand docs fix: feedback_arc_set() now disallows non-finite weights chore: update changelog feat: much faster exact minimum feedback arc set finding feat: igraph_find_cycle() feat: igraph_bitset_update() chore: update changelog [skip ci] feat: igraph_vector_reverse_section() and igraph_vector_rotate_left() feat: igraph_strvector_swap_elements() fix: the documented igraph_strvector_resize_min() was missing from headers fix: memory cleanup order in Pajek reader chore: code formatting [skip ci] fix: Pajek reader warns about duplicate vertex IDs, fixes igraph/igraph#2416 refactor: minor cleanup in DIMACS flow writer chore: update changelog [skip ci] refactor: rename vector_contains() parameter in header as well refactor: rename parameter in vector_contains() for consistency deprecate: `igraph_vector_binsearch2()` is deprecated in favour of `igraph_vector_contains_sorted()` deprecate: `igraph_vector_qsort_ind()` is deprecated in favour of `igraph_vector_sort_ind()`, closes igraph/igraph#2616 ci: more colour fix: remove macros referencing non-existent vector_list_t types, refs igraph/igraph#2662 refactor: clean up igraph_compose() test: test igraph_malloc() test: memory allocation tests doc: clean up memory allocation docs, fix non-included section refactor: make variables more local in UMAP fix: restore functionality of UMAP_DEBUG chore: update changelog fix: add missing error check in UMAP fix: properly center UMAP layout in 3D refactor: readability cleanup in UMAP code fix: crash in igraph_layout_umap() when passing distances == NULL and distance_are_weights == true refactor: adjlist_has_edge / adjlist_replace_edge cleanup refactor: igraph_set_t formatting nitpicks refactor: better error message in rewire() refactor: more cleanup in fr layout fuzzing: add misc_algos_weighted fuzzer which was already on the develop branch refactor: readability and header fix in umap implementation refactor: readability in kk and fr layout implementations doc: doc improvements for chung_lu and watts_strogatz chore: fix typos in changelog [skip ci] doc: document vector_is_equal() ref igraph/igraph#2653 deprecate: igraph_array3_t refactor: config header cleanup ci: attempt to fix manual branch selection for coverity run again test: fix memory leak in igraph_modularity() test ci: fix manual selection of branch for coverity workflow fix: memory leak introduced with previous fixes chore: update changelog chore: fix typo in parameter name fix: add missing error check fix: null-dereference in community_optimal_modularity() when passing the null or singleton graph, requesting modularity, but not requesting membership fix: null-dereference in igraph_community_voronoi() when membership == NULL and modularity != NULL fix: igraph_modularity() no longer allows NULL for 'modularity' as this is pointless and wasn't working correctly anyway doc: documentation nitpicks and standardization chore: update changelog [skip ci] fix: work around potential infinite loop in igraph_community_multilevel() by repeated re-randomization of the vertex processing order fixes igraph/igraph#2650 fuzzer: use libxml2 2.13.2 [skip ci] doc: fix typos in shortest paths docs chore: update changelog [skip ci] bench: fix memory leak in igraph_tree_game() benchmark bench: fix memory leak in induced_subgraph_edges() benchmark bench: minor cleanup and copyright headers bench: add Leiden, Louvain community detection benchmark chore: fix typo in changelog [skip ci] chore: fix copyright headers in bitset source files doc: improve bitset docs chore: update changelog [skip ci] doc: update old reference to deprecated function doc: some improved doc formatting chore: stylistic nitpick [skip ci] doc: fix misformatting in bitset docs that prevented the inclusion of a doc line doc: remove mention of long-remove stack_ptr_t ci: move coverity over to develop refactor: deprecate igraph_minimum_spanning_tree_(prim|unweighted)() chore: correct Zenodo DOI chore: Delete CODE_OF_CONDUCT.md (igraph/igraph#2645) doc: commnity detection doc polish doc: fix typo [skip ci] interface: add default variant for Floyd-Warshall chore: fix typo in docs [skip ci]

…vertex_set

chore: improve changelog fix: eigenvector centralization now always assumes scaled vertex-level centrality scores doc: more cross-linking for harmonic centrality doc: add reference to betweenness docs ci: apt-get update to work around outdated index ci: test with clang 20 doc: doc cleanup in centralization functions refactor: minor cleanup / IDE-friendliness for community_to_membership() ci: re-enable travis on ppc64 / s390x

szhorvat · 2024-09-19T11:23:18Z

Thanks @krlmlr! I am almost surprised that the last C core change didn't cause revdepcheck failures, but I won't complain about good news :-)

szhorvat marked this pull request as ready for review August 22, 2024 09:19

szhorvat commented Aug 22, 2024

View reviewed changes

R/aaa-auto.R Outdated Show resolved Hide resolved

szhorvat commented Aug 22, 2024

View reviewed changes

szhorvat force-pushed the try/c-core-2024-08-22 branch from 259abce to fa9dca7 Compare August 22, 2024 17:30

szhorvat changed the title ~~chore: Update vendored sources to igraph/igraph@d37be6e5a28bc6c299b4656313b7255c8378c17c~~ chore: Update vendored sources to igraph/igraph@03760a09cc7195dfaad3a1e60cff1632aa6a6118 Aug 22, 2024

krlmlr force-pushed the try/c-core-2024-08-22 branch from fa9dca7 to f68f630 Compare August 22, 2024 20:03

krlmlr closed this Aug 22, 2024

krlmlr reopened this Aug 22, 2024

krlmlr force-pushed the try/c-core-2024-08-22 branch from f68f630 to 9733b8d Compare August 22, 2024 20:04

krlmlr closed this Aug 22, 2024

krlmlr reopened this Aug 22, 2024

szhorvat force-pushed the try/c-core-2024-08-22 branch from c264b1e to c282139 Compare August 24, 2024 20:17

krlmlr force-pushed the try/c-core-2024-08-22 branch from c282139 to 20262b1 Compare September 1, 2024 03:28

schochastics mentioned this pull request Sep 6, 2024

problem with new igraph version termehs/netropy#3

Closed

maelle added the hackathon-topic 📆 label Sep 17, 2024

krlmlr force-pushed the try/c-core-2024-08-22 branch from b2360ea to f4b1736 Compare September 18, 2024 04:53

krlmlr added the mergequeue label Sep 18, 2024

aviator-app bot added the blocked label Sep 18, 2024

krlmlr force-pushed the try/c-core-2024-08-22 branch from d74de1b to c26ab44 Compare September 18, 2024 18:58

szhorvat and others added 5 commits September 18, 2024 20:59

refactor: disable code generation for all distances functions since t…

9d6c1b1

…hese are exposed manually, and this is unlikely to change for a while

chore: sync functions-R.yaml to C core, temporarily disable feedback_…

b5c7250

…vertex_set

Results

d4b6899

krlmlr force-pushed the try/c-core-2024-08-22 branch from c26ab44 to d4b6899 Compare September 18, 2024 18:59

krlmlr removed blocked hackathon-topic 📆 labels Sep 18, 2024

aviator-app bot merged commit c07844d into main Sep 18, 2024
11 checks passed

aviator-app bot deleted the try/c-core-2024-08-22 branch September 18, 2024 20:02

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

chore: Update vendored sources to igraph/igraph@03760a09cc7195dfaad3a1e60cff1632aa6a6118 #1470

chore: Update vendored sources to igraph/igraph@03760a09cc7195dfaad3a1e60cff1632aa6a6118 #1470

szhorvat commented Aug 22, 2024 •

edited

Loading

aviator-app bot commented Aug 22, 2024 •

edited

Loading

szhorvat commented Aug 22, 2024 •

edited

Loading

szhorvat Aug 22, 2024 •

edited

Loading

szhorvat commented Aug 22, 2024

Antonov548 commented Aug 22, 2024

szhorvat commented Aug 22, 2024

krlmlr commented Aug 22, 2024 •

edited

Loading

krlmlr commented Aug 22, 2024

szhorvat commented Aug 24, 2024

krlmlr commented Aug 29, 2024

szhorvat commented Aug 31, 2024

krlmlr commented Sep 1, 2024

szhorvat commented Sep 1, 2024

szhorvat commented Sep 5, 2024

szhorvat commented Sep 5, 2024

krlmlr commented Sep 5, 2024

szhorvat commented Sep 5, 2024 •

edited

Loading

szhorvat commented Sep 5, 2024

maelle commented Sep 5, 2024

krlmlr commented Sep 5, 2024

krlmlr commented Sep 5, 2024

schochastics commented Sep 6, 2024

schochastics commented Sep 6, 2024

szhorvat commented Sep 6, 2024

termehs commented Sep 6, 2024

szhorvat commented Sep 12, 2024

szhorvat commented Sep 12, 2024 •

edited

Loading

szhorvat commented Sep 17, 2024

krlmlr commented Sep 18, 2024

aviator-app bot commented Sep 18, 2024

krlmlr commented Sep 18, 2024

szhorvat commented Sep 19, 2024

chore: Update vendored sources to igraph/igraph@03760a09cc7195dfaad3a1e60cff1632aa6a6118 #1470

chore: Update vendored sources to igraph/igraph@03760a09cc7195dfaad3a1e60cff1632aa6a6118 #1470

Conversation

szhorvat commented Aug 22, 2024 • edited Loading

aviator-app bot commented Aug 22, 2024 • edited Loading

Current Aviator status

szhorvat commented Aug 22, 2024 • edited Loading

szhorvat Aug 22, 2024 • edited Loading

Choose a reason for hiding this comment

szhorvat commented Aug 22, 2024

Antonov548 commented Aug 22, 2024

szhorvat commented Aug 22, 2024

krlmlr commented Aug 22, 2024 • edited Loading

krlmlr commented Aug 22, 2024

szhorvat commented Aug 24, 2024

krlmlr commented Aug 29, 2024

szhorvat commented Aug 31, 2024

krlmlr commented Sep 1, 2024

szhorvat commented Sep 1, 2024

szhorvat commented Sep 5, 2024

szhorvat commented Sep 5, 2024

krlmlr commented Sep 5, 2024

szhorvat commented Sep 5, 2024 • edited Loading

manynet

MetaNet

netropy

priorCON

R6causal

simcausal

skynet

szhorvat commented Sep 5, 2024

maelle commented Sep 5, 2024

krlmlr commented Sep 5, 2024

krlmlr commented Sep 5, 2024

schochastics commented Sep 6, 2024

schochastics commented Sep 6, 2024

szhorvat commented Sep 6, 2024

termehs commented Sep 6, 2024

szhorvat commented Sep 12, 2024

szhorvat commented Sep 12, 2024 • edited Loading

szhorvat commented Sep 17, 2024

krlmlr commented Sep 18, 2024

aviator-app bot commented Sep 18, 2024

krlmlr commented Sep 18, 2024

szhorvat commented Sep 19, 2024

szhorvat commented Aug 22, 2024 •

edited

Loading

aviator-app bot commented Aug 22, 2024 •

edited

Loading

szhorvat commented Aug 22, 2024 •

edited

Loading

szhorvat Aug 22, 2024 •

edited

Loading

krlmlr commented Aug 22, 2024 •

edited

Loading

szhorvat commented Sep 5, 2024 •

edited

Loading

szhorvat commented Sep 12, 2024 •

edited

Loading