Bug#956925: dpkg: dpkg-source: should fail before-build if patches can't be applied fully in v3-quilt

Tomas Janousek <[email protected]>
Newsgroups gmane.linux.debian.devel.dpkg.bugs
Message-ID <Z9w4m-c16edNgYoP__8283.69481475733$1742486433$gmane$org@notes.lisk.in>
Hello,

On Mon, May 11, 2020 at 01:40:42AM +0200, Jiri Palecek wrote:
>OK. I have thought about those other workflows and it should be possible
>to support it while maintaining a sane function for people using quilt.
>My assumption is that when you have the whole tree in git and do not
>store .pc, as is discussed in bug 680155, dpkg should not apply the
>patches, therefore never create the .pc directory. This can be used to
>distinguish these users to users with .pc metadata tracking applied
>patches. Of course the first patch heuristic is imprecise (as 680155
>shows), but it's been good enough till now so we can go along with that.
>
>The attached patch just checks that the .pc directory exists and if it
>doesn't, applies the heuristic. If it exist, I assume the info in the
>.pc directory should be good enough to get applied patches list from.
>The patch contains a test that checks if it works under both scenarios
>(you need to have quilt installed to test it fully).

Is there going to be any .pc directory when you (or buildd) build the 
package from scratch for the first time? I suspect not, so there's still 
the possiblity that patches will be silently skipped whenever the first 
one no longer applies cleanly (such as when you update to a new upstream 
version).

Or am I missing something?

-- 
Tomáš "liskin" ("Pivník") Janoušek, https://lisk.in/
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.