* Re: Why unrelated package trigger rebuild?
@ 2020-10-09 16:26 zimoun
0 siblings, 0 replies; 4+ messages in thread
From: zimoun @ 2020-10-09 16:26 UTC (permalink / raw)
To: Guix Devel
Hi Chris,
I have investigate a bit.
On Fri, 9 Oct 2020 at 09:52, Christopher Baines <mail@cbaines.net> wrote:
> > <https://data.guix.gnu.org/repository/1/branch/master/package/ghc-haddock/output-history>
The webpage reads, anti-chronological order:
j5llsz…-ghc-haddock-2.22.0 392220 1a2759
s7s3ks…-ghc-haddock-2.22.0 20cc8f 20cc8f
lw2qzd…-ghc-haddock-2.22.0 8ab70b c1f385
where the log is:
--8<---------------cut here---------------start------------->8---
$ git --no-pager guix log --oneline 20cc8f9487^^..3922208091
3922208091 gnu: Add dkgpg.
d51cb7f85f gnu: Add libtmcg.
[...]
12dc9f58c4 gnu: gcc: Fix libstdc++ misconfiguration.
b1d3c122e1 gnu: gcc@10: Update to 10.2.0.
c1f38577c8 gnu: Add sleef.
548d4d810d gnu: duplicity: Update to 0.8.14.
[...]
e239135c56 gnu: gp2c: Update to 0.0.11pl4.
7cce2a6bd4 Revert "gnu: patchelf: Update to 0.11."
20cc8f9487 gnu: patchelf: Update to 0.11.
dbdfbec957 gnu: git-annex: Update to 8.20200720.1.
--8<---------------cut here---------------end--------------->8---
Well, it is probable that there is a merge and maybe the issue is coming
from that. Or maybe an author vs commit date. WDYT?
What I get is:
--8<---------------cut here---------------start------------->8---
3922208091 /gnu/store/j5llszq1cf12qn79bvhc042wc06ibfix-ghc-haddock-2.22.0
d51cb7f85f /gnu/store/j5llszq1cf12qn79bvhc042wc06ibfix-ghc-haddock-2.22.0
12dc9f58c4 /gnu/store/j5llszq1cf12qn79bvhc042wc06ibfix-ghc-haddock-2.22.0
b1d3c122e1 /gnu/store/lw2qzdrx5sq9qrpdszx4fby05bbdhi5f-ghc-haddock-2.22.0
c1f38577c8 /gnu/store/lw2qzdrx5sq9qrpdszx4fby05bbdhi5f-ghc-haddock-2.22.0
548d4d810d /gnu/store/lw2qzdrx5sq9qrpdszx4fby05bbdhi5f-ghc-haddock-2.22.0
e239135c56 /gnu/store/lw2qzdrx5sq9qrpdszx4fby05bbdhi5f-ghc-haddock-2.22.0
7cce2a6bd4 /gnu/store/lw2qzdrx5sq9qrpdszx4fby05bbdhi5f-ghc-haddock-2.22.0
20cc8f9487 /gnu/store/s7s3ksfhkdbb6k6si8bjnxy8vvywv322-ghc-haddock-2.22.0
dbdfbec957 /gnu/store/lw2qzdrx5sq9qrpdszx4fby05bbdhi5f-ghc-haddock-2.22.0
--8<---------------cut here---------------end--------------->8---
Therefore, the first commit producing j5llsz…-ghc-haddock is not
3922208091 but 12dc9f58c4. But what does it mean “first”? :-)
author date commit date
3922208091|Thu Jul 23 14:03:19 2020|Fri Jul 24 14:21:31 2020
12dc9f58c4|Fri Jul 24 09:56:25 2020|Fri Jul 24 14:21:30 2020
However, it seems more expected that
12dc9f58c4 gnu: gcc: Fix libstdc++ misconfiguration.
introduces new hash to ’ghc-haddock’ than unrelated
3922208091 gnu: Add dkgpg.
Maybe a possible fix is to compare “commit date” which will be more
accurate about the change (especially with merges) than “author
date”. WDYT?
The s7s3ks…-ghc-haddock is explained by the revert so it is expected.
All the best,
simon
where the script:
--8<---------------cut here---------------start------------->8---
for ci in 3922208091 \
d51cb7f85f \
12dc9f58c4 \
b1d3c122e1 \
c1f38577c8 \
548d4d810d \
e239135c56 \
7cce2a6bd4 \
20cc8f9487 \
dbdfbec957
do
printf "$ci "
guix time-machine --commit=$ci \
-- build ghc-haddock \
2> /dev/null
done
--8<---------------cut here---------------end--------------->8---
^ permalink raw reply [flat|nested] 4+ messages in thread
* Why unrelated package trigger rebuild?
@ 2020-10-09 1:35 zimoun
2020-10-09 7:51 ` Christopher Baines
2020-11-18 8:45 ` zimoun
0 siblings, 2 replies; 4+ messages in thread
From: zimoun @ 2020-10-09 1:35 UTC (permalink / raw)
To: Guix Devel
Dear,
From the Data Service:
<https://data.guix.gnu.org/repository/1/branch/master/package/ghc-haddock/output-history>
the commit 39222080911eaf3d7f74effe4467c1a04464aef3 which is about the
addition of the package ’dkgpg’ modifies the hash of the package
’ghc-haddock’ from:
/gnu/store/s7s3ksfhkdbb6k6si8bjnxy8vvywv322-ghc-haddock-2.22.0
to:
/gnu/store/j5llszq1cf12qn79bvhc042wc06ibfix-ghc-haddock-2.22.0
Well, I miss why the addition of one package changes the hash of another
unrelated one? Is it expected?
All the best,
simon
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Why unrelated package trigger rebuild?
2020-10-09 1:35 zimoun
@ 2020-10-09 7:51 ` Christopher Baines
2020-11-18 8:45 ` zimoun
1 sibling, 0 replies; 4+ messages in thread
From: Christopher Baines @ 2020-10-09 7:51 UTC (permalink / raw)
To: zimoun; +Cc: guix-devel
zimoun <zimon.toutoune@gmail.com> writes:
> Dear,
>
> From the Data Service:
>
> <https://data.guix.gnu.org/repository/1/branch/master/package/ghc-haddock/output-history>
>
> the commit 39222080911eaf3d7f74effe4467c1a04464aef3 which is about the
> addition of the package ’dkgpg’ modifies the hash of the package
> ’ghc-haddock’ from:
>
> /gnu/store/s7s3ksfhkdbb6k6si8bjnxy8vvywv322-ghc-haddock-2.22.0
>
> to:
>
> /gnu/store/j5llszq1cf12qn79bvhc042wc06ibfix-ghc-haddock-2.22.0
>
> Well, I miss why the addition of one package changes the hash of another
> unrelated one? Is it expected?
I think the first thing to spot is that the Guix Data Service does a
poor job of showing what happened. If you look at the 3 outputs that
start or end in July (07), the pattern isn't a neat "one after the
other".
What I think actually happens is that:
- 2020-06-13 21:04:16: hash begins lw2
- 2020-07-21 21:57:03: hash begins s7s
- after 2020-07-21 21:57:03: hash goes back to beginning lw2
- 2020-07-24 12:22:11: hash begins j5l
I think one of the implicit assumptions the query for that page makes,
is that the outputs don't really repeat, and in this case, that doesn't
hold. I think there's room for improvement to try and correct that.
Sorry about that,
Chris
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Why unrelated package trigger rebuild?
2020-10-09 1:35 zimoun
2020-10-09 7:51 ` Christopher Baines
@ 2020-11-18 8:45 ` zimoun
1 sibling, 0 replies; 4+ messages in thread
From: zimoun @ 2020-11-18 8:45 UTC (permalink / raw)
To: Guix Devel
Hi,
For the interested reader, a quick cross-reference.
On Fri, 09 Oct 2020 at 03:35, zimoun <zimon.toutoune@gmail.com> wrote:
> From the Data Service:
>
> <https://data.guix.gnu.org/repository/1/branch/master/package/ghc-haddock/output-history>
>
> the commit 39222080911eaf3d7f74effe4467c1a04464aef3 which is about the
> addition of the package ’dkgpg’ modifies the hash of the package
> ’ghc-haddock’ from:
>
> /gnu/store/s7s3ksfhkdbb6k6si8bjnxy8vvywv322-ghc-haddock-2.22.0
>
> to:
>
> /gnu/store/j5llszq1cf12qn79bvhc042wc06ibfix-ghc-haddock-2.22.0
>
> Well, I miss why the addition of one package changes the hash of another
> unrelated one? Is it expected?
The correct answer is because the Data Service deals with “state”
message of the mailing list guix-commits. Where “state” means bunch of
commits pushed in a row. Therefore, somehow it is expected. However,
is it the behaviour we want is another question? :-)
Explanations are here:
<https://yhetil.org/guix-devel/863617oe1h.fsf@gmail.com>
and chat on IRC starting here:
<http://logs.guix.gnu.org/guix/2020-11-17.log#224837>
All the best,
simon
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2020-11-18 9:03 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-10-09 16:26 Why unrelated package trigger rebuild? zimoun
-- strict thread matches above, loose matches on Subject: below --
2020-10-09 1:35 zimoun
2020-10-09 7:51 ` Christopher Baines
2020-11-18 8:45 ` zimoun
Code repositories for project(s) associated with this public inbox
https://git.savannah.gnu.org/cgit/guix.git
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 read-only IMAP folder(s) and NNTP newsgroup(s).