Github messages for voidlinux
 help / color / mirror / Atom feed
* [PR PATCH] [RFC] gc: update to 8.2.0, cleanup.
@ 2021-11-20 22:48 tornaria
  2021-11-20 22:55 ` leahneukirchen
                   ` (7 more replies)
  0 siblings, 8 replies; 9+ messages in thread
From: tornaria @ 2021-11-20 22:48 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 2158 bytes --]

There is a new pull request by tornaria against master on the void-packages repository

https://github.com/tornaria/void-packages gc
https://github.com/void-linux/void-packages/pull/34176

[RFC] gc: update to 8.2.0, cleanup.
I was looking for out of date dependencies for sage so I updated `gc` to 8.2.0. After I had already finished I realized that 8.2.0 is experimental; it seems all the `*.*.0` releases are experimental so I added an ignore to the update file.

Even if this is not to merge now (although it might be useful to merge the ignore), I made some cleanup for some things that seem to be old and not needed anymore -- but since I'm not sure I'm asking for comments.

Cleanup 1: for musl, the current template sets `CFLAGS='-D_GNU_SOURCE -DNO_GETCONTEXT -DSEARCH_FOR_DATA_START -DUSE_MMAP -DHAVE_DL_ITERATE_PHDR'` but all of this seems noop except maybe `-DSEARCH_FOR_DATA_START` which I don't understand; it seems originally came from alpine which afaict doesn't use it anymore. In any case I'd argue for using the same CFLAGS for all archs and since this seems to be working without any...

Cleanup 2: there is a `post_install()` which installed headers, in fact it just copied all of the `include` dir in the source which also contains build artifacts. It seems to me the upstream make install works ok. It should be pointed out that some headers are not installed, seemingly those related to c++. But if c++ support is needed, passing `--enable-cplusplus` to configure seems the right way and then it will not only install those headers but also install a couple extra shared libraries so maybe installing just the .h is useless.

My test of the resulting packages (after the cleanup) is to compile and check ecl (using #34175) and while a small number of tests fail it doesn't seem to be the change in gc causing that (I tested on glibc and musl with both 8.0.6 and 8.2.0).

Maybe @ivmai can give us some feedback about these changes, and also please confirm that `*.*.0` is always experimental so it's safe to ignore in the update-check.

A patch file from https://github.com/void-linux/void-packages/pull/34176.patch is attached

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: github-pr-gc-34176.patch --]
[-- Type: text/x-diff, Size: 1680 bytes --]

From 76f3f48e3a5e7e571d17fdd0662c9ae1a7ac4096 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Gonzalo=20Tornar=C3=ADa?= <tornaria@cmat.edu.uy>
Date: Sat, 20 Nov 2021 18:56:44 -0300
Subject: [PATCH] gc: update to 8.2.0, cleanup.

---
 srcpkgs/gc/template | 16 ++--------------
 srcpkgs/gc/update   |  1 +
 2 files changed, 3 insertions(+), 14 deletions(-)
 create mode 100644 srcpkgs/gc/update

diff --git a/srcpkgs/gc/template b/srcpkgs/gc/template
index 3e00fa2ab468..a4a2834289cf 100644
--- a/srcpkgs/gc/template
+++ b/srcpkgs/gc/template
@@ -1,6 +1,6 @@
 # Template file for 'gc'
 pkgname=gc
-version=8.0.6
+version=8.2.0
 revision=1
 build_style=gnu-configure
 configure_args="--enable-static --enable-mmap"
@@ -11,19 +11,7 @@ maintainer="Orphaned <orphan@voidlinux.org>"
 license="GPL-2.0-or-later"
 homepage="https://www.hboehm.info/gc/"
 distfiles="https://github.com/ivmai/bdwgc/releases/download/v${version}/gc-${version}.tar.gz"
-checksum=3b4914abc9fa76593596773e4da671d7ed4d5390e3d46fbf2e5f155e121bea11
-
-case "$XBPS_TARGET_MACHINE" in
-*-musl)
-	CFLAGS='-D_GNU_SOURCE -DNO_GETCONTEXT -DSEARCH_FOR_DATA_START -DUSE_MMAP -DHAVE_DL_ITERATE_PHDR'
-esac
-
-post_install() {
-	mkdir -p ${DESTDIR}/usr/include/gc/
-	mv ${wrksrc}/include/* ${DESTDIR}/usr/include/gc/
-	mv ${DESTDIR}/usr/include/gc/extra/* ${DESTDIR}/usr/include/
-	rmdir ${DESTDIR}/usr/include/gc/extra
-}
+checksum=2540f7356cb74f6c5b75326c6d38a066edd796361fd7d4ed26e494d9856fed8f
 
 gc-devel_package() {
 	depends="gc>=${version}_${revision}"
diff --git a/srcpkgs/gc/update b/srcpkgs/gc/update
new file mode 100644
index 000000000000..05aaeed314e6
--- /dev/null
+++ b/srcpkgs/gc/update
@@ -0,0 +1 @@
+ignore=*.*.0

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [RFC] gc: update to 8.2.0, cleanup.
  2021-11-20 22:48 [PR PATCH] [RFC] gc: update to 8.2.0, cleanup tornaria
@ 2021-11-20 22:55 ` leahneukirchen
  2021-11-21  8:08 ` ivmai
                   ` (6 subsequent siblings)
  7 siblings, 0 replies; 9+ messages in thread
From: leahneukirchen @ 2021-11-20 22:55 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 182 bytes --]

New comment by leahneukirchen on void-packages repository

https://github.com/void-linux/void-packages/pull/34176#issuecomment-974723046

Comment:
sgtm, but i'd just wait for 8.2.1.

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [RFC] gc: update to 8.2.0, cleanup.
  2021-11-20 22:48 [PR PATCH] [RFC] gc: update to 8.2.0, cleanup tornaria
  2021-11-20 22:55 ` leahneukirchen
@ 2021-11-21  8:08 ` ivmai
  2021-11-21  8:21 ` ivmai
                   ` (5 subsequent siblings)
  7 siblings, 0 replies; 9+ messages in thread
From: ivmai @ 2021-11-21  8:08 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 550 bytes --]

New comment by ivmai on void-packages repository

https://github.com/void-linux/void-packages/pull/34176#issuecomment-974772717

Comment:
Hi @tornaria,

> After I had already finished I realized that 8.2.0 is experimental; it seems all the *.*.0 releases are experimental so I added an ignore to the update file.

Yes, all *.*.0 releases are experimental (they are marked as Pre-release in Github releases). I'm working on stabilization (but the progress is no so fast as I expected, most probably stable release will merge before Feb 2022).



^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [RFC] gc: update to 8.2.0, cleanup.
  2021-11-20 22:48 [PR PATCH] [RFC] gc: update to 8.2.0, cleanup tornaria
  2021-11-20 22:55 ` leahneukirchen
  2021-11-21  8:08 ` ivmai
@ 2021-11-21  8:21 ` ivmai
  2021-11-21  8:30 ` ivmai
                   ` (4 subsequent siblings)
  7 siblings, 0 replies; 9+ messages in thread
From: ivmai @ 2021-11-21  8:21 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1265 bytes --]

New comment by ivmai on void-packages repository

https://github.com/void-linux/void-packages/pull/34176#issuecomment-974774146

Comment:
> Cleanup 1: for musl, the current template sets CFLAGS='-D_GNU_SOURCE -DNO_GETCONTEXT -DSEARCH_FOR_DATA_START -DUSE_MMAP -DHAVE_DL_ITERATE_PHDR' but all of this seems noop except maybe -DSEARCH_FOR_DATA_START which I don't understand; it seems originally came from alpine which in fact doesn't use it anymore. In any case I'd argue for using the same CFLAGS for all archs and since this seems to be working without any...

Yes, I think most of them are no-op as you listed. As for SEARCH_FOR_DATA_START, I have several regular musl jobs on Travis (see (.travis.yml)[https://github.com/ivmai/bdwgc/blob/master/.travis.yml]) w/o this macro defined explicitly - the tests always pass. So, I don't know whether it is needed or not in practice. It look like both approaches are reliable - rely on __data_start or scanning of write-accessible space down from _end in case SEARCH_FOR_DATA_START, or rely on __environ otherwise (the default in upstream).

Up to you but my suggestion: to remove SEARCH_FOR_DATA_START and see if there any bug reports (or search for similar bugs in the past) , if yes then report the bug upstream.

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [RFC] gc: update to 8.2.0, cleanup.
  2021-11-20 22:48 [PR PATCH] [RFC] gc: update to 8.2.0, cleanup tornaria
                   ` (2 preceding siblings ...)
  2021-11-21  8:21 ` ivmai
@ 2021-11-21  8:30 ` ivmai
  2021-11-21  8:38 ` ivmai
                   ` (3 subsequent siblings)
  7 siblings, 0 replies; 9+ messages in thread
From: ivmai @ 2021-11-21  8:30 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 454 bytes --]

New comment by ivmai on void-packages repository

https://github.com/void-linux/void-packages/pull/34176#issuecomment-974775397

Comment:
> Cleanup 2: there is a post_install() which installed headers, in fact it just copied all of the include dir in the source which also contains build artifacts. It seems to me the upstream make install works ok. 

Sure. I just haven't tried this in my (PR)[https://github.com/void-linux/void-packages/pull/33426].

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [RFC] gc: update to 8.2.0, cleanup.
  2021-11-20 22:48 [PR PATCH] [RFC] gc: update to 8.2.0, cleanup tornaria
                   ` (3 preceding siblings ...)
  2021-11-21  8:30 ` ivmai
@ 2021-11-21  8:38 ` ivmai
  2021-11-21 12:55 ` [PR PATCH] [Closed]: " tornaria
                   ` (2 subsequent siblings)
  7 siblings, 0 replies; 9+ messages in thread
From: ivmai @ 2021-11-21  8:38 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 766 bytes --]

New comment by ivmai on void-packages repository

https://github.com/void-linux/void-packages/pull/34176#issuecomment-974776386

Comment:
> My test of the resulting packages (after the cleanup) is to compile and check ecl (using #34175) and while a small number of tests fail it doesn't seem to be the change in gc causing that (I tested on glibc and musl with both 8.0.6 and 8.2.0

Sure. The experimental status of the release is a message to the downstream package maintainers and other users "a release with new features and significant code changes is prepared, a lot of testing has been done by the upstream maintainer on some major targets like linux/x64, please test it in your environments and/or give your users ability to test this experimental release.

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [PR PATCH] [Closed]: [RFC] gc: update to 8.2.0, cleanup.
  2021-11-20 22:48 [PR PATCH] [RFC] gc: update to 8.2.0, cleanup tornaria
                   ` (4 preceding siblings ...)
  2021-11-21  8:38 ` ivmai
@ 2021-11-21 12:55 ` tornaria
  2021-11-21 12:55 ` tornaria
  2021-11-21 15:26 ` ivmai
  7 siblings, 0 replies; 9+ messages in thread
From: tornaria @ 2021-11-21 12:55 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 2010 bytes --]

There's a closed pull request on the void-packages repository

[RFC] gc: update to 8.2.0, cleanup.
https://github.com/void-linux/void-packages/pull/34176

Description:
I was looking for out of date dependencies for sage so I updated `gc` to 8.2.0. After I had already finished I realized that 8.2.0 is experimental; it seems all the `*.*.0` releases are experimental so I added an ignore to the update file.

Even if this is not to merge now (although it might be useful to merge the ignore), I made some cleanup for some things that seem to be old and not needed anymore -- but since I'm not sure I'm asking for comments.

Cleanup 1: for musl, the current template sets `CFLAGS='-D_GNU_SOURCE -DNO_GETCONTEXT -DSEARCH_FOR_DATA_START -DUSE_MMAP -DHAVE_DL_ITERATE_PHDR'` but all of this seems noop except maybe `-DSEARCH_FOR_DATA_START` which I don't understand; it seems originally came from alpine which afaict doesn't use it anymore. In any case I'd argue for using the same CFLAGS for all archs and since this seems to be working without any...

Cleanup 2: there is a `post_install()` which installed headers, in fact it just copied all of the `include` dir in the source which also contains build artifacts. It seems to me the upstream make install works ok. It should be pointed out that some headers are not installed, seemingly those related to c++. But if c++ support is needed, passing `--enable-cplusplus` to configure seems the right way and then it will not only install those headers but also install a couple extra shared libraries so maybe installing just the .h is useless.

My test of the resulting packages (after the cleanup) is to compile and check ecl (using #34175) and while a small number of tests fail it doesn't seem to be the change in gc causing that (I tested on glibc and musl with both 8.0.6 and 8.2.0).

Maybe @ivmai can give us some feedback about these changes, and also please confirm that `*.*.0` is always experimental so it's safe to ignore in the update-check.

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [RFC] gc: update to 8.2.0, cleanup.
  2021-11-20 22:48 [PR PATCH] [RFC] gc: update to 8.2.0, cleanup tornaria
                   ` (5 preceding siblings ...)
  2021-11-21 12:55 ` [PR PATCH] [Closed]: " tornaria
@ 2021-11-21 12:55 ` tornaria
  2021-11-21 15:26 ` ivmai
  7 siblings, 0 replies; 9+ messages in thread
From: tornaria @ 2021-11-21 12:55 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 596 bytes --]

New comment by tornaria on void-packages repository

https://github.com/void-linux/void-packages/pull/34176#issuecomment-974812710

Comment:
@ivmai: thanks a lot for your comments. Since void linux means to ship only stable releases, we'll skip 8.2.0. I think when a new stable version is released we can proceed with the cleanups outlined in this PR (and also add `*.*.0` to ignore so we know to skip those releases in the future).

Regarding `SEARCH_FOR_DATA_START` and musl, just a note that  that alpine is not currently using it (https://git.alpinelinux.org/aports/tree/main/gc/APKBUILD).

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [RFC] gc: update to 8.2.0, cleanup.
  2021-11-20 22:48 [PR PATCH] [RFC] gc: update to 8.2.0, cleanup tornaria
                   ` (6 preceding siblings ...)
  2021-11-21 12:55 ` tornaria
@ 2021-11-21 15:26 ` ivmai
  7 siblings, 0 replies; 9+ messages in thread
From: ivmai @ 2021-11-21 15:26 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 258 bytes --]

New comment by ivmai on void-packages repository

https://github.com/void-linux/void-packages/pull/34176#issuecomment-974837437

Comment:
> Regarding SEARCH_FOR_DATA_START and musl, just a note that that alpine is not currently using it

Got it. Thank you

^ permalink raw reply	[flat|nested] 9+ messages in thread

end of thread, other threads:[~2021-11-21 15:26 UTC | newest]

Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-11-20 22:48 [PR PATCH] [RFC] gc: update to 8.2.0, cleanup tornaria
2021-11-20 22:55 ` leahneukirchen
2021-11-21  8:08 ` ivmai
2021-11-21  8:21 ` ivmai
2021-11-21  8:30 ` ivmai
2021-11-21  8:38 ` ivmai
2021-11-21 12:55 ` [PR PATCH] [Closed]: " tornaria
2021-11-21 12:55 ` tornaria
2021-11-21 15:26 ` ivmai

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).