[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#1026886: marked as done (fakeupstream.cgi: support functionality to ignore pre-release on GitHub.com)

Your message dated Fri, 21 Apr 2023 20:06:34 +0900
with message-id <20230421200626.1aca0952@jet>
and subject line Re: watch file for packages on github
has caused the Debian Bug report #1026886,
regarding fakeupstream.cgi: support functionality to ignore pre-release on GitHub.com
to be marked as done.

This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.

(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact owner@bugs.debian.org
immediately.)


-- 
1026886: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1026886
Debian Bug Tracking System
Contact owner@bugs.debian.org with problems
--- Begin Message --- 
Package: qa.debian.org
Severity: normal
X-Debbugs-Cc: kenhys@xdump.org

# Problem

Currently, it seems that there is no comprehensive way to
ignore pre-release on GitHub.com [1]

Adding a feature upstream=github_releases to avoid unexpected failure
to do packaging pre-release as an official release.

# Failure scenario

Here is the failure scenario:

  1. upstream make a tag as v1.0.0
  2. upstream releases v1.0.0 as pre-release
  3. upstream remove a tag v1.0.0
  4. upstream make a tag v1.0.0 again
  5. upstream releases v1.0.0 as official release

If you overlooked step 2 as a pre-release, d/watch
just notify v1.0.0 released and shows it in
package overview [2]

As a result, it may cause wrong version packaged into archive.
If so, you need to apply many patch sets against to v1.0.0-2.

These kind of failure is derived from difficulty of
distinguish (with uscan) pre-release from
https://github.com/org/project/releases or
https://github.com/org/project/tags

Practically, it is a rare case, so I'm not sure that it should be fixed or
redirector is used frequently, but I'll file it as an example of rare case.

# Expected

If redirector(fakeupstream.cgi) returns a response which excludes
pre-release explicitly by default, it may improves such a situation.

Usage in debian/watch (assumption):

  version=4
  opts="searchmode=plain" \
    http://qa.local/cgi-
bin/fakeupstream.cgi?upstream=github_releases/google/sentencepiece \
    .*/sentencepiece-@ANY_VERSION@@ARCHIVE_EXT@

# Additional Information

NOTE: The API response [3] vary whether assets are attached or not.

With assets, it returns browser_download_url field:
https://github.com/(org)/(project)/releases/download/(tag_name)/{project}-{version}.tar.gz

Without assets, it returns zipball_url or tarball_url fields:
  https://api.github.com/repos/(org)/(project)/tarball/(tag_name)
  https://api.github.com/repos/(org)/(project)/zipball/(tag_name)

[1] debian/watch: Ignoring pre-release on GitHub
    https://lists.debian.org/debian-devel/2022/12/msg00024.html

[2] https://qa.debian.org/developer.php?login=[your account]

[3] https://api.github.com/repos/$org/$project/releases

--- End Message ---
--- Begin Message --- 
On Fri, 21 Apr 2023 19:21:35 +0900 Kentaro Hayashi <kenhys@xdump.org> wrote:
> Control: tags -1 wontfix
> 
> > 
> > Anyway, it may be a trivial exceptional case, so I'll close it.
> 

closed.

--- End Message ---
Reply to: