Regen #81

EPashkin · 2018-03-01T15:25:28Z

Also contains deignoring string constants

EPashkin · 2018-03-01T15:27:20Z

@GuillaumeGomez, @sdroege, @tmiasko
I not sure about deignoring constants. Maybe its really needed ignored as used in C printing functions.

GuillaumeGomez · 2018-03-01T15:31:43Z

gdk-sys/src/lib.rs

@@ -3675,7 +3675,7 @@ pub struct GdkEventTouchpadPinch {
    pub type_: GdkEventType,
    pub window: *mut GdkWindow,
    pub send_event: i8,
-    pub phase: GdkTouchpadGesturePhase,


Is it normal?

EPashkin · 2018-03-01T15:34:23Z

Yes, you merged patch to gir-files today, see gtk-rs/gir#555 (comment) for reasoning

GuillaumeGomez · 2018-03-01T15:36:00Z

Indeed, then all good for me!

sdroege · 2018-03-02T08:45:34Z

glib-sys/src/lib.rs

-//pub const G_GUINT64_FORMAT: *const c_char = b"lu\0" as *const u8 as *const c_char;
-//pub const G_GUINTPTR_FORMAT: *const c_char = b"lu\0" as *const u8 as *const c_char;
+pub const G_GUINT64_FORMAT: *const c_char = b"lu\0" as *const u8 as *const c_char;
+pub const G_GUINTPTR_FORMAT: *const c_char = b"lu\0" as *const u8 as *const c_char;


These constants are all useless in Rust. They only make sense with printf-style API, but fortunately we have a better format string system in Rust

sdroege · 2018-03-02T08:46:01Z

glib-sys/src/lib.rs

+pub const G_GSIZE_FORMAT: *const c_char = b"lu\0" as *const u8 as *const c_char;
+pub const G_GSIZE_MODIFIER: *const c_char = b"l\0" as *const u8 as *const c_char;
+pub const G_GSSIZE_FORMAT: *const c_char = b"li\0" as *const u8 as *const c_char;
+pub const G_GSSIZE_MODIFIER: *const c_char = b"l\0" as *const u8 as *const c_char;
 pub const G_GUINT16_FORMAT: *const c_char = b"hu\0" as *const u8 as *const c_char;
 pub const G_GUINT32_FORMAT: *const c_char = b"u\0" as *const u8 as *const c_char;


I.e. even these existing ones were useless.

sdroege · 2018-03-02T08:46:24Z

glib-sys/src/lib.rs

 pub const G_GUINT16_FORMAT: *const c_char = b"hu\0" as *const u8 as *const c_char;
 pub const G_GUINT32_FORMAT: *const c_char = b"u\0" as *const u8 as *const c_char;
-//pub const G_GUINT64_FORMAT: *const c_char = b"lu\0" as *const u8 as *const c_char;
-//pub const G_GUINTPTR_FORMAT: *const c_char = b"lu\0" as *const u8 as *const c_char;
+pub const G_GUINT64_FORMAT: *const c_char = b"lu\0" as *const u8 as *const c_char;


And this one for example is platform dependent. Both CPU and compiler (MSVC again)

EPashkin · 2018-03-02T15:14:44Z

Then I ignore it again

EPashkin · 2018-03-09T06:20:06Z

Ignored more constants
Added fixes for GPid and GPollFD

Still don't contains update to right gir submodule.

EPashkin · 2018-03-09T06:37:45Z

Appveyor failed on 64x build as glib have some GPid functions.

sdroege · 2018-03-09T08:31:44Z

glib-sys/src/lib.rs

-//pub const G_GINT64_FORMAT: *const c_char = b"li\0" as *const u8 as *const c_char;
-//pub const G_GINT64_MODIFIER: *const c_char = b"l\0" as *const u8 as *const c_char;
-//pub const G_GINTPTR_FORMAT: *const c_char = b"li\0" as *const u8 as *const c_char;
-//pub const G_GINTPTR_MODIFIER: *const c_char = b"l\0" as *const u8 as *const c_char;
 pub const G_GNUC_FUNCTION: *const c_char = b"\0" as *const u8 as *const c_char;
 pub const G_GNUC_PRETTY_FUNCTION: *const c_char = b"\0" as *const u8 as *const c_char;


These are also both useless, needs a C compiler to do something

sdroege · 2018-03-09T08:31:53Z

glib-sys/src/lib.rs

@@ -648,24 +651,8 @@ pub const G_DATE_BAD_YEAR: c_int = 0;
 pub const G_DIR_SEPARATOR: c_int = 92;
 pub const G_DIR_SEPARATOR_S: *const c_char = b"\\\0" as *const u8 as *const c_char;


These two are platform specific (hello Windows)

Yes, I just ignored it wrong.

sdroege · 2018-03-09T08:32:12Z

glib-sys/src/lib.rs

-pub const G_GUINT16_FORMAT: *const c_char = b"hu\0" as *const u8 as *const c_char;
-pub const G_GUINT32_FORMAT: *const c_char = b"u\0" as *const u8 as *const c_char;
-//pub const G_GUINT64_FORMAT: *const c_char = b"lu\0" as *const u8 as *const c_char;
-//pub const G_GUINTPTR_FORMAT: *const c_char = b"lu\0" as *const u8 as *const c_char;
 pub const G_HAVE_GINT64: c_int = 1;
 pub const G_HAVE_GNUC_VARARGS: c_int = 1;
 pub const G_HAVE_GNUC_VISIBILITY: c_int = 1;


These 3 are useless outside C too

sdroege · 2018-03-09T08:32:18Z

glib-sys/src/lib.rs

@@ -720,10 +707,6 @@ pub const G_MAXUINT32: u32 = 4294967295;
 pub const G_MAXUINT64: u64 = 18446744073709551615;
 pub const G_MAXUINT8: u8 = 255;


as are these

sdroege · 2018-03-09T08:32:33Z

glib-sys/src/lib.rs

-//pub const G_MININT16: i16 = -32768;
-//pub const G_MININT32: i32 = -2147483648;
-pub const G_MININT64: i64 = -9223372036854775808;
-//pub const G_MININT8: i8 = -128;
 pub const GLIB_MINOR_VERSION: c_int = 54;
 pub const G_MODULE_SUFFIX: *const c_char = b"so\0" as *const u8 as *const c_char;


This one is platform specific

sdroege · 2018-03-09T08:32:40Z

glib-sys/src/lib.rs

@@ -740,10 +723,6 @@ pub const G_PRIORITY_HIGH_IDLE: c_int = 100;
 pub const G_PRIORITY_LOW: c_int = 300;
 pub const G_SEARCHPATH_SEPARATOR: c_int = 59;
 pub const G_SEARCHPATH_SEPARATOR_S: *const c_char = b";\0" as *const u8 as *const c_char;


Platform specific

IMHO is not

It's ; on Windows and : everywhere else

Hm, really.
Strange that it have constant name="SEARCHPATH_SEPARATOR_S" value=";" c:type="G_SEARCHPATH_SEPARATOR_S in ubuntu GLib-2.0.gir.

That is a bug in g-ir-scanner. It tries to parse macro definition without actually implementing a proper C preprocessing step. If macros have conditional definitions then it is quite likely to give wrong results.

It would be nice to extend ABI checker to constants as well. I would expect to find a lot of incorrect things there.

sdroege · 2018-03-09T08:33:24Z

glib-sys/src/manual.rs

+use libc::{c_int, c_ushort};
+
+#[cfg(all(windows,target_arch="x86_64"))]
+pub type GPid = i64; //Actually HANDLE aka void*


Shouldn't this be a *mut c_void then?

It depend of #82 discussion

sdroege · 2018-03-09T08:34:05Z

Basically most of the GLib constants are not very useful. Opt-in instead of opt-out for each of them would be more convenient I guess

EPashkin · 2018-03-09T09:10:54Z

Bad that we can't Opt-in for constants in sys mode

EPashkin · 2018-03-09T09:35:42Z

Fixed almost all nits.

EPashkin · 2018-03-09T09:51:36Z

Almost final version, only may have changes from glib problem.

EPashkin · 2018-03-11T10:10:32Z

I restarted Appveyor.

sdroege · 2018-03-11T10:29:14Z

glib-sys/src/lib.rs

-//pub const G_GUINTPTR_FORMAT: *const c_char = b"lu\0" as *const u8 as *const c_char;
-pub const G_HAVE_GINT64: c_int = 1;
-pub const G_HAVE_GNUC_VARARGS: c_int = 1;
-pub const G_HAVE_GNUC_VISIBILITY: c_int = 1;
 pub const G_HAVE_GROWING_STACK: c_int = 0;
 pub const G_HAVE_ISO_VARARGS: c_int = 1;


Should go away, these two. basically all G_HAVE_.* and G_.*FORMAT and G_.*MODIFIER

sdroege · 2018-03-11T10:29:32Z

glib-sys/src/lib.rs

-//pub const G_MININT16: i16 = -32768;
-//pub const G_MININT32: i32 = -2147483648;
-pub const G_MININT64: i64 = -9223372036854775808;
-//pub const G_MININT8: i8 = -128;
 pub const GLIB_MINOR_VERSION: c_int = 54;


As written in another ticket somewhere, these are kind of useless too

sdroege · 2018-03-11T10:30:12Z

glib-sys/src/manual.rs

+pub type GPid = *mut c_void;
+
+#[cfg(not(windows))]
+pub type GPid = c_int;


sdroege · 2018-03-11T10:30:18Z

glib-sys/src/manual.rs

+pub struct GPollFD {
+    pub fd: c_int,
+    pub events: c_ushort,
+    pub revents: c_ushort,


EPashkin · 2018-03-11T11:26:55Z

Removed version constants in glib and in others.

EPashkin · 2018-03-13T15:55:21Z

@GuillaumeGomez IMHO this ready for merge now

GuillaumeGomez · 2018-03-13T16:04:43Z

Indeed, thanks!

GuillaumeGomez reviewed Mar 1, 2018

View reviewed changes

sdroege reviewed Mar 2, 2018

View reviewed changes

EPashkin force-pushed the regen branch from 7e3319b to e7db8b6 Compare March 2, 2018 15:14

EPashkin force-pushed the regen branch from e7db8b6 to c035f6a Compare March 9, 2018 06:18

This was referenced Mar 9, 2018

GPid and GPollFD ABI mismatch for Windows #82

Closed

Use GObject status when generating types gtk-rs/gir#563

Merged

sdroege reviewed Mar 9, 2018

View reviewed changes

EPashkin force-pushed the regen branch from c035f6a to fafec06 Compare March 9, 2018 09:34

EPashkin force-pushed the regen branch from fafec06 to 97a69b2 Compare March 9, 2018 09:43

Update submodules

5dc3254

EPashkin force-pushed the regen branch from 97a69b2 to e99f40b Compare March 9, 2018 09:50

EPashkin force-pushed the regen branch from e99f40b to a71ca1a Compare March 9, 2018 10:00

EPashkin force-pushed the regen branch from a71ca1a to 4d388e8 Compare March 9, 2018 10:25

This was referenced Mar 9, 2018

Regen gtk-rs/glib#296

Merged

Cross validate Rust ABI with C ABI. gtk-rs/gir#558

Merged

ghost mentioned this pull request Mar 11, 2018

Cross validate the ABI of sys crates. gtk-rs/gir#555

Closed

sdroege reviewed Mar 11, 2018

View reviewed changes

glib-sys/src/manual.rs

pub type GPid = *mut c_void;

#[cfg(not(windows))]

pub type GPid = c_int;

Copy link

Member

sdroege Mar 11, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

👍

sdroege reviewed Mar 11, 2018

View reviewed changes

glib-sys/src/manual.rs

pub struct GPollFD {

pub fd: c_int,

pub events: c_ushort,

pub revents: c_ushort,

Copy link

Member

sdroege Mar 11, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

👍

EPashkin force-pushed the regen branch 2 times, most recently from bcf5421 to d826521 Compare March 11, 2018 10:45

EPashkin added 3 commits March 11, 2018 14:02

Ignore all C related or not portable constants

35e8be7

Make GPid and GPollFD manual

7a8e39c

Regen

1285a1d

EPashkin force-pushed the regen branch from d826521 to 1285a1d Compare March 11, 2018 11:26

GuillaumeGomez merged commit e01a0e0 into gtk-rs:master Mar 13, 2018

EPashkin deleted the regen branch March 13, 2018 16:08

		@@ -648,24 +651,8 @@ pub const G_DATE_BAD_YEAR: c_int = 0;
		pub const G_DIR_SEPARATOR: c_int = 92;
		pub const G_DIR_SEPARATOR_S: const c_char = b"\\\0" as const u8 as *const c_char;

		@@ -720,10 +707,6 @@ pub const G_MAXUINT32: u32 = 4294967295;
		pub const G_MAXUINT64: u64 = 18446744073709551615;
		pub const G_MAXUINT8: u8 = 255;

Regen #81

Regen #81

Conversation

EPashkin commented Mar 1, 2018

EPashkin commented Mar 1, 2018

Choose a reason for hiding this comment

EPashkin commented Mar 1, 2018

GuillaumeGomez commented Mar 1, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

EPashkin commented Mar 2, 2018

EPashkin commented Mar 9, 2018

EPashkin commented Mar 9, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sdroege commented Mar 9, 2018

EPashkin commented Mar 9, 2018

EPashkin commented Mar 9, 2018

EPashkin commented Mar 9, 2018

EPashkin commented Mar 11, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

EPashkin commented Mar 11, 2018

EPashkin commented Mar 13, 2018

GuillaumeGomez commented Mar 13, 2018