From: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
To: buildroot@busybox.net
Subject: [Buildroot] [PATCH 0/4] Sanetize packages version
Date: Wed, 12 Jun 2019 09:26:24 +0200 [thread overview]
Message-ID: <20190612092624.1ca7836d@windsurf> (raw)
In-Reply-To: <20190612064209.23619-1-victor.huesca@bootlin.com>
Hello,
On Wed, 12 Jun 2019 08:42:05 +0200
Victor Huesca <victor.huesca@bootlin.com> wrote:
> In particular packages fetched via git cannot be easily sanitized. The
> implementation behind the git method is to use the <pkg>_VERSION after the
> rope cloning to checkout the version tag. A workaround could be to add a
> new variable to the infrastructure that can be set to specify the tag for
> these cases.
I guess an example of this is the "at" package, which has:
AT_VERSION = release/3.1.23
AT_SITE = https://salsa.debian.org/debian/at.git
AT_SITE_METHOD = git
but we would probably want the version to be just "3.1.23", like is
tracked at https://release-monitoring.org/project/127/.
So indeed, I see two possible directions here:
(1) Keep the <pkg>_VERSION semantic as it is today, and add a separate
<pkg>_UPSTREAM_VERSION variable (or some other name) that holds the
upstream version. This would give:
AT_UPSTREAM_VERSION = 3.1.23
AT_VERSION = release/$(AT_UPSTREAM_VERSION)
With this, no need to change anything in the package
infrastructure, we just need <pkg>_UPSTREAM_VERSION be extracted
by support/scripts/pkg-stats, which is can trivially do instead of
extracting <pkg>_VERSION.
This approach also has the advantage that the name of the tarballs
stored on disk doesn't change, which means all the DL_DIR contents
+ sources.b.o content remain identical.
Of course, the package infrastructure defines
<pkg>_UPSTREAM_VERSION to <pkg>_VERSION if there's no
<pkg>_UPSTREAM_VERSION value defined for a given package.
(2) As suggested by Victor, add a separate variable that tells the Git
download logic what is the actual tag/version to fetch, i.e:
AT_VERSION = 3.1.23
AT_GIT_COMMIT = release/$(AT_UPSTREAM_VERSION)
Perhaps option (1) is the easiest ? But it has the drawback that
<pkg>_VERSION doesn't contain just the actual version, but like we do
today, some possibly semi-random stuff in addition to the version (like
"release/1.2.3" or "foo-1.2.3).
> Also, there are a few packages that use the <pkg>_VERSION variable to store
> more than just the version. For example gap-madam-bin-maxi have a suffix
I don't see a gap-madam-bin-maxi package in Buildroot. I am missing
something ?
> depending on whereas the frame buffer is selected. An other example is lua-
> coatpersistent which seems to have two backends and store this in the
> version variable.
>
> A last case I found is the case of Kodi. Kodi's major versions have a
> codename in addition to the version number (eg. 18.2 "Leia" or 17.3 "Krypton").
> This codename may or not be consider as an inherent part of the version. I
> would tend to consider this as part of the version and keep it as a version
> suffix for all kodi-related packages as it is currently.
Kodi has two different entries in release-monitoring.org:
https://release-monitoring.org/project/5511/, which has only the
version, but seems to be outdated
https://release-monitoring.org/project/20547/, which has the version
+ codename, and seems to be updated
So it looks like we should use the second one, and therefore keep the
code name ?
Best regards,
Thomas
--
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
next prev parent reply other threads:[~2019-06-12 7:26 UTC|newest]
Thread overview: 29+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-06-12 6:42 [Buildroot] [PATCH 0/4] Sanetize packages version Victor Huesca
2019-06-12 6:42 ` [Buildroot] [PATCH 1/4] package: remove 'v' prefix from github-fetched packages Victor Huesca
2019-06-19 20:50 ` Arnout Vandecappelle
2019-06-20 6:31 ` Thomas Petazzoni
2019-06-20 19:33 ` Arnout Vandecappelle
2019-06-20 19:46 ` Thomas Petazzoni
2019-06-20 21:07 ` Arnout Vandecappelle
2019-06-20 11:51 ` Victor Huesca
2019-06-20 19:31 ` Arnout Vandecappelle
2019-06-12 6:42 ` [Buildroot] [PATCH 2/4] package: remove 'v' prefix from tarball-fetched packages Victor Huesca
2019-06-19 21:06 ` Arnout Vandecappelle
2019-06-20 6:13 ` Thomas Petazzoni
2019-06-20 9:23 ` Victor Huesca
2019-06-20 19:43 ` Arnout Vandecappelle
2019-06-12 6:42 ` [Buildroot] [PATCH 3/4] package: remove non-conventional prefix/suffix from github-fetched packages Victor Huesca
2019-06-19 21:30 ` Arnout Vandecappelle
2019-06-20 12:42 ` Victor Huesca
2019-06-20 19:50 ` Arnout Vandecappelle
2019-06-21 2:15 ` Carlos Santos
2019-06-12 6:42 ` [Buildroot] [PATCH 4/4] package: remove non-conventional prefix/suffix from tarball-fetched packages Victor Huesca
2019-06-20 21:27 ` Arnout Vandecappelle
2019-06-12 7:26 ` Thomas Petazzoni [this message]
2019-06-12 8:39 ` [Buildroot] [PATCH 0/4] Sanetize packages version Victor Huesca
2019-06-12 14:50 ` Thomas Petazzoni
2019-06-19 15:35 ` Victor Huesca
2019-06-20 20:32 ` Arnout Vandecappelle
2019-06-23 16:33 ` Thomas Petazzoni
2019-06-12 8:51 ` Arnout Vandecappelle
2019-06-18 12:11 ` Victor Huesca
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20190612092624.1ca7836d@windsurf \
--to=thomas.petazzoni@bootlin.com \
--cc=buildroot@busybox.net \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox