public inbox for isar-users@googlegroups.com
 help / color / mirror / Atom feed
From: Uladzimir Bely <ubely@ilbers.de>
To: isar-users@googlegroups.com, Jan Kiszka <jan.kiszka@siemens.com>
Subject: Re: [PATCH] sbuild: Don't update apt database on every package build
Date: Wed, 26 Oct 2022 10:06:01 +0300	[thread overview]
Message-ID: <2564476.vYhyI6sBWr@hp> (raw)
In-Reply-To: <08331fcb-e592-8ee1-70c7-072932657528@siemens.com>

In mail from вторник, 25 октября 2022 г. 16:37:51 +03 user Jan Kiszka wrote:
> On 25.10.22 11:00, Uladzimir Bely wrote:
> > In mail from понедельник, 24 октября 2022 г. 17:14:52 +03 user Jan Kiszka
> > 
> > wrote:
> >> On 24.10.22 15:10, Uladzimir Bely wrote:
> >>> By default, sbuild updates apt database beforethe build. It may cause
> >>> an inconsistency between apt databases for different packages built at
> >>> the different time if external (Debian) mirrors has updated recently.
> >>> 
> >>> With new changes, sbuild will rely on global apt state that was got
> >>> during debootstrap stage.
> >>> 
> >>> Local `isar-apt` repo will still be automatically updated while it is
> >>> passed via `--extra-repository=` argument.
> >> 
> >> Will that also ensure upgrades, e.g. from essential debian packages to
> >> self-built ones hosted in isar-apt? We had a few of such cases in the
> >> past already in downstream layers, and I recall permitting upgrades to
> >> the old buildchroot for that.
> > 
> > I'm afraid I don't exactly get you question...
> > 
> > Do you mean, it may happen that Isay may update 'isar-apt' repository, but
> > won't even try to install "newer" (custom) package from isar-apt as an
> > upgrade of some Debian package installed at debootstrap stage?
> 
> Actually, part of my former problems were downgrades, not upgrades, see
> [1] and [2]. Those are possible as well because anything coming from
> isar-apt has higher prior than other repo. Now I was only wondering if
> that behavior is still given, with all those switches applied and
> schroot in between us and the actual apt calls.

It seems sbuild man pages say nothing about "--allow-downgrades". So, I'm not 
sure if it's used.

Anyway, now I guess that setting `--no-apt-upgrade` and `--no-apt-distupgrade` 
in the v1 of patch is wrong. Most probably, such configuration won't allow 
even upgrades from debian packages to isar-apt ones.

Also, you are saying, "isar-apt has higher prior than other repo". Where is it 
configured? I thought, `isar-apt` is just one of sources from "apt" and if 
Debian has newer package version than we compiled by hand, these newer 
dependencies are to be installed...

> 
> > Yes, probably it may happen. The case I tried to cover is just to make
> > sbuild know about custom packages in 'isar-apt' so it could install them
> > as depends. But if the package already installed (e.g. at debootstrap or
> > sbuild-chroot stage) it may happen it won't be upgraded.
> > 
> > I guess, there should be someting like "--no-apt-update --apt-distupgrade"
> > then. So, apt database won't be updated, but if 'isar-apt' provides new
> > version of already installed (Debian) package, it will be used.
> > 
> > Or, even "--no-apt-update" should be enough, while "$apt_distupgrade" is
> > already set by default in sbuild...
> > 
> > 
> > Additinally, it would be good to have some test for this case. For
> > example, we could have "hello" (from Debian) preinstalled before build,
> > but upgrade it to custom "patched" version during sbuild.
> 
> Yes, that would indeed be good.
> 
> Jan
> 
> [1]
> https://github.com/ilbers/isar/commit/5367ba74d9f2f0b82560bab6c146f0521e1304
> 9f [2]
> https://github.com/ilbers/isar/commit/77e62c1e9bbda107a10136d1aa5f26f5f66603
> 71





  reply	other threads:[~2022-10-26  7:06 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-10-24 13:10 Uladzimir Bely
2022-10-24 14:14 ` Jan Kiszka
2022-10-25  9:00   ` Uladzimir Bely
2022-10-25 13:37     ` Jan Kiszka
2022-10-26  7:06       ` Uladzimir Bely [this message]
2022-10-26  7:17         ` Jan Kiszka

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=2564476.vYhyI6sBWr@hp \
    --to=ubely@ilbers.de \
    --cc=isar-users@googlegroups.com \
    --cc=jan.kiszka@siemens.com \
    /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