| Commit message (Collapse) | Author | Age |
... | |
|
|
|
|
|
| |
No functional change with the existing callers.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
|
|
|
|
|
| |
This was archive_api_query_curl. We're working on making it good for
other things too.
No functional change.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
|
|
|
|
|
|
| |
We don't want a global SIGPIPE setting. In particular we do not want
to run any of subprocesses with SIGPIPE ignored.
Just in case libcurl gets this wrong, tell it CURLOPT_NOSIGNAL, and do
it ourselves.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
There is a bug in libdpkg-perl, where Dpkg::Source::Package installs a
handler on loading. This has the effect of making in-process C
calls (eg to libcurl) uninterruptible by ^C.
Work around it by unconditionally resetting SIGINT. (Anyone who
deliberately invoked us with SIGINT ignored deserves to keep all the
resulting pieces.)
See #932841.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
| |
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
| |
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
|
|
|
|
|
|
|
| |
I don't think anyone is using these. We are going to do something
different to try fo fix #932570.
When #790093 is fixed we may need something like archive-query-tls-key
(or even exactly that) but we would have to reimplement it for
WWW::Curl anyway.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
|
|
|
| |
The principal effect in the Debian configuration is to access the
master dgit git server via ssh, rather than the mirror via https.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
|
|
|
| |
This is much more conveient and in some cases sensible than
-cdgit-distro.DISTRO.readonly=0
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
|
|
|
|
|
|
|
|
| |
In support of #932570. Now there is only one caller of
archive_api_query_cmd.
It is OK to pass `undef' because
- archive_
No functional change.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
|
|
|
| |
In support of #932570. Our intended other caller doesn't have a
$data. No functional change.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
|
|
| |
In support of #932570. No functional change.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
| |
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
| |
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
|
| |
|
|
|
|
| |
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
|
|
|
|
|
|
| |
Mention two omitted bug Closes.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
| |
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
| |
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
|
|
|
|
|
|
|
| |
See discussion in #932612.
Suggested-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
|
|
|
|
|
|
| |
Closes: #932612
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
|
|
|
|
|
|
| |
That is mostly what we are using now.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
| |
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
| |
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
|
|
|
|
|
|
| |
Closes: #932477
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
|
|
|
|
| |
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
|
|
|
|
|
|
|
| |
Otherwise, when using git-debpush in a terminal, the user will get a
pile of unwanted diff output each time check_treesame is called.
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
|
|
|
|
| |
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
|
|
|
|
| |
When looking for -sn or -sk in debian/source/options, we need to not
print it to stdout even if grep finds it. So use -q.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
Acked-by: Sean Whitton <spwhitton@spwhitton.name>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This means that a run with set -x shows the tag data that is being
signed. The user can even then try it again. It will make it easier
to do some kind of dry run, too, if we want to.
We are really not in danger of hitting a command line length limit.
(It also involves fewer syscalls, less use of /tmp etc., although that
is really irrelevant here.)
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
Acked-by: Sean Whitton <spwhitton@spwhitton.name>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This was a normal git remote pointing to the dgit git server.
This is not really sensible because
1. it can't simply be pushed to
2. sometimes (if the package was never pushed with dgit)
it doesn't even exist
3. it may be out of date with respect to the archive
The effect of 2 is that it can break `git remote update'.
This "git remote add" was introduced in 2013 in
39c6c123ab730d42ec7c9ed01c30e0175c7691e7
fix branch usage
and was first included in 0.2. There doesn't seem to be any evidence
of me having deeply considered the merits of this at the time.
Closes: #932694
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
|
|
|
|
|
|
| |
Our distro is test-dummy. We are going to stop randomly fetching tags
from the dgit git server, so the debian/ one would be omitted.
Replace the tag with a fresh equivalent one and change the one use
site.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
|
|
|
| |
The name is from when I thought it would actually mkdir. But it
doesn't. No overall functional change.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
|
| |
Reported-by: Sean Whitton <spwhitton@spwhitton.name>
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
|
|
|
|
|
| |
I work for the Wellcome Sanger Institute, but am employed by Genome
Research Limited, so it is they who should be credited in Copyright
notices.
Signed-off-by: Matthew Vernon <mv3@sanger.ac.uk>
Closes: #932630
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Nicer short description right in the first section.
- Don't use the term 'delta queue' until it's needed.
- Don't talk about other manpages so early.
- Cater for people who type `man git-debrebase` looking for a
barebones tutorial, not (just) a command line reference.
- That tutorial should minimise references to dgit, especially since
users might want to use git-debpush to upload.
Thanks to Sam Hartman for the user feedback which prompted this and
Ian Jackson for feedback on the content of "QUICK REFERENCE".
Closes: #926656
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
|
|
|
|
|
|
|
| |
This was there so that dgit's -c options would be honoured here.
But now, we get this information via @git, which dgit adds -c
options to, which we pass to git-playtree-create by invoking it
via git.
So this is no longer used and can be abolished.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
|
|
|
|
| |
Not used yet, so no overall functional change to the script.
The .deb has a new file.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
|
|
|
|
|
|
|
|
|
|
| |
We want this because git-debpush is going to want this functionality
but doesn't want to include (or have a copy of) Dgit.pm.
The installation arrangements are rather ugly.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
|
|
|
|
|
|
|
|
|
| |
This clears the way for some parts of Dgit.pm to honour dgit's
git invocation conventions.
No functional change.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
| |
|
|
|
|
|
|
| |
Closes: #932476
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
Acked-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
| |
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
|
| |
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
Closes: #857490
|
|
|
|
|
| |
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
Closes: #932536
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
While using git-debpush interactively, if the user sees that checks
have failed which they know to be safe to override, they typically
just use --force or -f.
However, it is useful for scripts to be able to skip single checks,
and if the user knows that a check will fail in advance of running
git-debpush, it is convenient to be able to specify that but still see
whether other checks fail, where those failures were unexpected. This
can avoid the user having to run git-debpush more than once.
The list of any checks which failed but were overridden is not stored
in the generated tag. git-debpush's checks are for the convenience of
the local user only, and the list of failed but overridden checks is
not considered to be metadata for the upload. Not recording failed
but overridden checks in the git tag keeps git-debpush's checking
independent of its wrapping of git-tag and git-push, which makes
git-debpush simpler and easier to understand. The complexity is on
the server side, in tag2upload.
git-debpush(1): We want the list of checks that can be overridden to
be at the bottom of the list of options because most users will not
need to look at it. We also want the description of --force|-f to be
adjacent to the description of --force=<check>. So move the
description of --force|-f to the end of the list of options.
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
Closes: #932459
|
|
|
|
|
|
| |
No functional change.
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
|
|
|
|
|
|
| |
Checks are still performed, but turned from errors into warnings.
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
|
|
|
|
| |
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|
|
|
|
| |
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
|