Opinions Wanted: Which Slackware Package Manager for an SBo SlackBuild Developer ( and why )?

fourtysixandtwo · 10-03-2022, 05:54 AM

kjhambrick,

Correction, intent belies the truth...tried 5GB, 5GB + 4GB swap, and then right to 10GB. I only boot to runlevel 3 too.

Just finished building pandoc and all the deps successfully.

Code:

$ pandoc -v
pandoc 2.17.1.1
Compiled with pandoc-types 1.22.1, texmath 0.12.4, skylighting 0.12.2,
citeproc 0.6.0.1, ipynb 0.2

edit:
Tried just building ghc with only 8GB of ram and it hit OOM again.

Code:

... kernel: oom-kill:constraint=CONSTRAINT_NONE,nodemask=(null),cpuset=/,mems_allowed=0,global_oom,task_memcg=/,task=ghc-stage1,pid=31608,uid=0

kjhambrick · 10-03-2022, 09:03 AM

Thank you fourtysixandtwo !

I'll give it a shot after work !

-- kjh

p.s. Thanks for lesson:

Code:

# grep oom-kill /var/log/messages*

messages.1:Sep 25 18:29:48 slack15 kernel: [  pid  ]   uid  tgid total_vm      rss pgtables_bytes swapents oom_score_adj name
messages.1:Sep 25 18:29:48 slack15 kernel: oom-kill:constraint=CONSTRAINT_NONE,nodemask=(null),cpuset=/,mems_allowed=0,global_oom,task_memcg=/,task=ghc-stage1,pid=28214,uid=0

Code:

# grep -e oom-kill -e 28214 /var/log/messages*

messages.1:Sep 25 18:29:48 slack15 kernel: [  pid  ]   uid  tgid total_vm      rss pgtables_bytes swapents oom_score_adj name
messages.1:Sep 25 18:29:48 slack15 kernel: [  28214]     0 28214 268508766   405851  7413760   497056             0 ghc-stage1
messages.1:Sep 25 18:29:48 slack15 kernel: oom-kill:constraint=CONSTRAINT_NONE,nodemask=null),cpuset=/,mems_allowed=0,global_oom,task_memcg=/,task=ghc-stage1,pid=28214,uid=0

kjhambrick · 10-03-2022, 04:43 PM

OK then.

Here's one that I've seen before but I forget which SBo SlackBuild it was ...

I've always set LC_ALL=C so that my ancient Scripts work ( mostly for sorting ).

However the haskell-lpeg.SlackBuild does not like that.

If I run sbopkg with the following settings:

Code:

export LC_ALL=en_US.UTF-8
export LANG=en_US.UTF-8 
export LANGUAGE=en_US.UTF-8

then haskell-lpeg.SlackBuild works !

So I've set my LANG and LC_ Variables as above and I started the pandoc Queue again

-- kjh

This is the haskell-lpeg.SlackBuild log with my default LANGUAGE and LC_ Settings.

It FAILS:

Code:

# head -9  haskell-lpeg.SlackBuild-1.0.2.log ; echo -e '\n<<snip>>\n' ; tail -4 haskell-lpeg.SlackBuild-1.0.2.log

haskell-lpeg.SlackBuild startup  at Mon Oct  3 16:19:04 CDT 2022
# 
# my Locale via:  env |grep -i -e lang -e lc_
#
LANG=en_US.UTF-8
LC_ALL=C
LC_COLLATE=C
#
lpeg-1.0.2/
lpeg-1.0.2/CHANGELOG.md

<<snip>>

Preprocessing library for lpeg-1.0.2..
Running Haddock on library for lpeg-1.0.2..
haddock: internal error: dist/doc/html/lpeg/haddock-response40900-1.txt: hGetContents: invalid argument (invalid byte sequence)

haskell-lpeg.SlackBuild FAILED   at Mon Oct  3 16:19:08 CDT 2022 ... RetCode = 1

OTOH, if I force a UTF8 Environment then haskell-lpeg.SlackBuild works !

Code:

# head -9  haskell-lpeg.SlackBuild-1.0.2_SBo_env.log ; echo -e '\n<<snip>>\n' ; tail -4 haskell-lpeg.SlackBuild-1.0.2_SBo_env.log

haskell-lpeg.SlackBuild startup  at Mon Oct  3 16:19:10 CDT 2022
# 
# my Locale via:  env |grep -i -e lang -e lc_
#
LANGUAGE=en_US.UTF-8
LANG=en_US.UTF-8
LC_ALL=en_US.UTF-8
LC_COLLATE=C
#
lpeg-1.0.2/

<<snip>>

Slackware package /tmp/haskell-lpeg-1.0.2-x86_64-1_SBo_env.tgz created.

haskell-lpeg.SlackBuild complete at Mon Oct  3 16:19:14 CDT 2022 ... RetCode = 0

kjhambrick · 10-03-2022, 05:50 PM

Woo Hoo !

Thanks fourtysixandtwo !!

I was able to compile pandoc via sbopkg using the unadulterated sqg queue.

Attached are all the MD5SUM Errors.

I answered 'R' to each and everything seemed to work.

What causes them if Retry resolves the errors

Now that I know I need the UTF Language Settings, I'll reinstall Sphinx and redo the pandoc queue.

Maybe ghc Docs will build now ...

Thanks again !

-- kjh

fourtysixandtwo · 10-03-2022, 10:32 PM

Good to hear and welcome!

sbopkg sticks the downloads in /var/cache/sbopkg and creates links in the slackbuild directory. Works fine until you get downloads that share the same filename but are different. Retry will redownload the proper file. It's a bug but doesn't happen too often.

For example:

Code:

haskell# find ./ -ls|egrep cabal$ |sort -k13
  1450714      0 lrwxrwxrwx   1 root     root           25 Oct  3 00:35 ./haskell-aeson/1.cabal -> /var/cache/sbopkg/1.cabal
  1573255      0 lrwxrwxrwx   1 root     root           25 Oct  3 03:00 ./haskell-th-lift/1.cabal -> /var/cache/sbopkg/1.cabal
  1573300      0 lrwxrwxrwx   1 root     root           25 Oct  3 03:03 ./haskell-async/1.cabal -> /var/cache/sbopkg/1.cabal
  1573397      0 lrwxrwxrwx   1 root     root           25 Oct  3 00:32 ./haskell-uuid-types/1.cabal -> /var/cache/sbopkg/1.cabal
  1573517      0 lrwxrwxrwx   1 root     root           25 Oct  3 00:33 ./haskell-scientific/1.cabal -> /var/cache/sbopkg/1.cabal
   404943      0 lrwxrwxrwx   1 root     root           25 Oct  3 00:24 ./haskell-text-short/2.cabal -> /var/cache/sbopkg/2.cabal
  1573081      0 lrwxrwxrwx   1 root     root           25 Oct  2 11:38 ./haskell-data-fix/2.cabal -> /var/cache/sbopkg/2.cabal
  1573169      0 lrwxrwxrwx   1 root     root           25 Oct  3 03:03 ./haskell-unliftio-core/2.cabal -> /var/cache/sbopkg/2.cabal
  1573325      0 lrwxrwxrwx   1 root     root           25 Oct  3 00:25 ./haskell-semigroupoids/2.cabal -> /var/cache/sbopkg/2.cabal
  1584523      0 lrwxrwxrwx   1 root     root           25 Oct  3 00:30 ./haskell-witherable/2.cabal -> /var/cache/sbopkg/2.cabal
  1573209      0 lrwxrwxrwx   1 root     root           25 Oct  3 03:48 ./haskell-regex-pcre/3.cabal -> /var/cache/sbopkg/3.cabal
  1573327      0 lrwxrwxrwx   1 root     root           25 Oct  2 11:43 ./haskell-time-compat/3.cabal -> /var/cache/sbopkg/3.cabal
  1584514      0 lrwxrwxrwx   1 root     root           25 Oct  3 00:31 ./haskell-strict/3.cabal -> /var/cache/sbopkg/3.cabal
  1573160      0 lrwxrwxrwx   1 root     root           25 Oct  3 03:57 ./haskell-HsYAML/4.cabal -> /var/cache/sbopkg/4.cabal
   535803      0 lrwxrwxrwx   1 root     root           25 Oct  3 00:30 ./haskell-these/5.cabal -> /var/cache/sbopkg/5.cabal
   404944      0 lrwxrwxrwx   1 root     root           25 Oct  2 11:38 ./haskell-deepseq-generics/7.cabal -> /var/cache/sbopkg/7.cabal

Did a little digging on ghc and OOM. Looks like it might be a reocurring bug related to optimizations. https://gitlab.haskell.org/ghc/ghc/-/issues/21973

Changing -O2 to -O1 and it builds without issue with 8GB of memory at least.

edit:
Also wanted to mention that I use B. Watson's sbo-maintainer-tools slackbuild and his other sbostuff tools https://slackware.uk/~urchlay/repos/sbostuff

kjhambrick · 10-04-2022, 04:59 AM

Well, Dang.

More research to do ...

If I build and install the SBo Sphinx.SlackBuild on my system then build ghc then Sphinx fails to build the ghc Documentation on my System.

I was hoping that setting my LC_ and LANG to UTF8 would fix it.

Does anybody have any Ideas ?

Error is below my sig.

Edit: This issue was reported in Gentoo's Bugzilla – Bug 792558 along with a link in Comment #4:
ghc Upstream Patches. Will test the nine Patches and report.

-- kjh

Code:

chmod +x                                                     inplace/bin/runghc
/usr/bin/sphinx-build -b html -d docs/users_guide/.doctrees-html -w docs/users_guide/.log -n -D latex_paper_size=letter docs/users_guide docs/users_guide/build-html/users_guide
Running Sphinx v4.2.0

Configuration error:
There is a syntax error in your configuration file: (unicode error) 'unicodeescape' codec can't decode bytes in position 1-2: truncated \uXXXX escape (conf.py, line 110)

make[1]: *** [docs/users_guide/ghc.mk:16: docs/users_guide/build-html/users_guide/index.html] Error 2
make: *** [Makefile:128: all] Error 2

ghc:
Would you like to continue processing the rest of the
queue or would you like to abort?  If this failed
package is a dependency of another package in the queue
then it may not make sense to continue.

(Y)es to continue, (N)o to abort, (R)etry the build?: N

kjhambrick · 10-04-2022, 10:24 AM

OK then ...

I hand-edited the three applicable Files mentioned at ghc version 8.10.5 Documentation Patches and then generated Unified Diffs ( attached as ghc-8.10.4-docs-patches.tgz.txt ).

Then I modified the ghc.SlackBuild to apply the three patches and reinstalled the Sphinx.SlackBuild and then reran the patched SBo ghc.SlackBuild.

Everything, including the Docs, builds successfully now.

So, having done all that, ghc version 8.10.4 was released 5 Feb 2021 and may be too old to bootstrap a newer version of ghc ( latest versions are 9.4.2 or 9.2.4 ).

Anyhow, all I was really after was to test sbopkg with large queues -- sbopkg passes with flying colors !

Thanks to everyone for all the help !

-- kjh

I attached ghc-8.10.4-docs-patches.tgz as ghc-8.10.4-docs-patches.tgz.txt ...

Save the attached tar file as ghc-8.10.4-docs-patches.tgz and extract the contents into your ghc.SlackBuild Directory, then apply the patch below to the ghc.SlackBuild:
[code]
# cd /path/to/development/ghc
# tar -zxvf /path/to/ghc-8.10.4-docs-patches.tgz.txt

This is the Patch for the Version 8.10.4 ghc.SlackBuild to apply documentation patches so ghc.SlackBuild works when Sphinx is installed.

Code:

--- ghc.SlackBuild.orig 2022-03-11 15:05:40.937890704 -0600
+++ ghc.SlackBuild      2022-10-04 08:34:52.671379769 -0500
@@ -125,6 +125,10 @@
  \( -perm 666 -o -perm 664 -o -perm 640 -o -perm 600 -o -perm 444 \
   -o -perm 440 -o -perm 400 \) -exec chmod 644 {} \;
 
+patch --verbose -b -p1 < $CWD/patch-01-users_guide-conf.py
+patch --verbose -b -p1 < $CWD/patch-02-users_guide-expected-undocumented-flags.txt
+patch --verbose -b -p1 < $CWD/patch-03-users_guide-extending_ghc.rst
+
 GHC=$BOOTSTRAPGHC \
 CFLAGS="$SLKCFLAGS" \
 CXXFLAGS="$SLKCFLAGS" \