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

Re: gbp import-orig --uscan for the non-trivial cases



Hi Julien,

On Thu, Aug 28, 2025 at 05:49:38PM +0200, Julien Plissonneau Duquène wrote:
Your current watch file (as of a few minutes ago) is fine so don't change anything. What happened in your previous attempt is that you declared two conflicting sources for the same tarball:

Are those actually conflicting? I think that's a valid usecase.

What we have here is an upstream that releases irregularly, so that it was decided eventually to package "a git commit". I think it should be possible to pull a development snapshot into debian (for example, for expeimental) while still having a debian/watch that triggers on "official releases".

Having to change debian/watch and debian/gbp.conf accordingly is clumsy and unelegant. Shouldn't we have a more comfortable way to do that? I think that #1112329 also originated from my (mis)understanding that two stanzas in debian/watch are a valid usecase.

Le 2025-08-27 20:35, Marc Haber a écrit :

| Version: 5
| | Template: Github
| Owner: dlundquist
| Project: sniproxy
| | Source: https://github.com/dlundquist/sniproxy.git
| Matching-Pattern: HEAD
| Mode: git
| Filename-Mangle: s/^(@PACKAGE@)-0\.0/$1-0.6.2/

This appears in gbp's debug output as:

| uscan warn: more than one main upstream tarballs listed.

Okay, I didnt understand that warning properly.

| uscan warn: more than one main upstream tarballs listed. [Devscripts::Uscan::WatchSource: 614]

| Newest version of sniproxy on remote site is 0.6.2~git20240702.2d99653, local version is 0.6.1+git20250616
|  => Newer package available from:
|         => https://github.com/dlundquist/sniproxy.git HEAD
| uscan debug: line: download_file_and_sig()
| uscan debug: line: mkorigtargz()
| uscan debug: Keep git repo (../sniproxy-temporary.358789.git)

the new version was actually found, but no tarball was generated, because the download flag in uscan's code was reset by the first source not going on with a download. This could be considered a bug (not sure it's worth reporting or fixing though), or as some say "unspecified behaviour" resulting from the conflicting sources.

That would be a bug once the uscan authors decide to make it easier to switch back and forth between upstream development snapshots and upstream releases. Has this use case been on the radar in uscan development?

Hope this sheds some light on uscan mysteries.

It does. Appreciated.

Greetings
Marc

--
-----------------------------------------------------------------------------
Marc Haber         | "I don't trust Computers. They | Mailadresse im Header
Leimen, Germany    |  lose things."    Winona Ryder | Fon: *49 6224 1600402
Nordisch by Nature |  How to make an American Quilt | Fax: *49 6224 1600421


Reply to: