From: vijai kumar <vijaikumar.kanagarajan@gmail.com>
To: Henning Schild <henning.schild@siemens.com>
Cc: Vijai Kumar K <Vijaikumar_Kanagarajan@mentor.com>,
Baurzhan Ismagulov <ibr@radix50.net>,
Jan Kiszka <jan.kiszka@siemens.com>,
isar-users <isar-users@googlegroups.com>
Subject: Re: [PATCH v4 0/2] Custom ${S} series
Date: Thu, 1 Apr 2021 00:23:02 +0530 [thread overview]
Message-ID: <CALLGG_LrSW9cE156Ee00nFmKbDY6WbC3kb6PBd4N43N2OqBPyQ@mail.gmail.com> (raw)
In-Reply-To: <20210331103433.34d5414b@md1za8fc.ad001.siemens.net>
On Wed, Mar 31, 2021 at 2:04 PM Henning Schild
<henning.schild@siemens.com> wrote:
>
> Am Wed, 31 Mar 2021 13:07:17 +0530
> schrieb vijai kumar <vijaikumar.kanagarajan@gmail.com>:
>
> > Hi,
> >
> > On Wed, Mar 31, 2021 at 12:03 PM Vijai Kumar K
> > <Vijaikumar_Kanagarajan@mentor.com> wrote:
> > >
> > > Changes since V3:
> > > - Fix build issue with qemumipsel-stretch.
> > > (Need to pass --only source with --print-uris, since we are dealing
> > > with source package)
> >
> > This case is something interesting. If you see the hello package in
> > [1], There are 2 versions of the binary package.
> >
> > All x86 and ARM architectures have version "2.10-1+deb9u1"
> > Rest of the architectures like mips, s390 and ppc have "2.10-1+b1"
>
> That is weird on its own. But we probably had this problem even before
> this series.
> Do those new sources produce working packages for those other arches or
> do we really need to find the other sources?
It does produce a working package. At least for hello.
>
> > Since we are using --only-source option to download, apt skips the
> > binary to source translation and directly fetches the latest source.
> > Which is "2.10-1+deb9u1". That is how it could be done, since the
> > recipe here is just the debian source file name.
> > But if you see by apt-cache policy the installation candidate for
> > mipsel is 2.10-1+b1 and the corresponding source version is "2.10-1".
>
> Maybe we can also download that binary, maybe not store it, to make apt
> double check against the binary. Jan found preferences not working for
> apt-get source, maybe his case would also benefit from that. What i
> understand is that "--only-source" can maybe be skipped to download
> binary and source and get a cross check.
For that we need to know the binary package it generates. Not all
source packages produce binary
packages with the same name. I still believe using --only-source is
right, if the SRC_URI of the recipe is expected to point
to the source package name. Which is by the way is the case.
This also acts as a guard, in case the user incorrectly specifies the
binary package name in SRC_URI.
In such a case do_apt_fetch would simply fail.
>
> > Ideally we should pull 2.10-1 instead of 2.10-1+deb9u1 for mipsel.
> > That is what the binary is available for?
>
> We modify the package anyways, i guess it is important to produce a
> working package, not too much "taking old sources" because upstream did
> not build new ones for that arch.
I agree here.
Some other packages with similar scenarios are mentioned below(not a
complete list).
1. https://packages.debian.org/stretch/zstd
2. https://packages.debian.org/stretch/zsh-static
3. https://packages.debian.org/stretch/apt
4. https://packages.debian.org/stretch/busybox
The updated version comes in via the security feed. Where only x86 and arm based
binaries are uploaded.
This FAQ[1], explains why it is missing for certain archs.
I guess it should be okay to use the new source to generate the debs.
Maybe we could capture this
entire scenario in a doc. I could amend the PR to add some info in one
of the docs.
[1] https://www.debian.org/security/faq#archismissing
Thanks,
Vijai Kumar K
>
> Henning
>
> > I know a non-maintainer upload will have a different binary package
> > version (+bX), but will point to the same source.
> > Here there are 2 different sources for different archs.
> > I am not sure if there are other such packages which have a similar
> > scenario. Just wanted to gather info on it here.
> >
> > [1] https://packages.debian.org/stretch/hello
> >
> > Thanks,
> > Vijai Kumar K
> >
> > >
> > > CI build succeeded:
> > > http://ci.isar-build.org:8080/job/isar_vkk_devel/95/
> > >
> > > Changes since V2:
> > > - Introduce P2 to demonstrate apt:// without ${PV} set.
> > > - Address review comment from Henning.
> > > - Fix issue where do_apt_unpack is triggered for non apt://
> > > SRC_URIs.
> > >
> > > CI build going on:
> > > http://ci.isar-build.org:8080/job/isar_vkk_devel/93/
> > >
> > > Changes since V1:
> > >
> > > - Introduce a separate do_apt_unpack task
> > >
> > > Vijai Kumar K (2):
> > > dpkg-base: Introduce do_apt_unpack
> > > Modify hello to demonstrate apt:// without ${PV}
> > >
> > > .../recipes-app/hello/{hello.inc => hello.bb} | 0
> > > meta-isar/recipes-app/hello/hello_2.10.bb | 11 -------
> > > meta-isar/recipes-app/hello/hello_2.9.bb | 8 -----
> > > meta/classes/dpkg-base.bbclass | 29
> > > ++++++++++++++++--- 4 files changed, 25 insertions(+), 23
> > > deletions(-) rename meta-isar/recipes-app/hello/{hello.inc =>
> > > hello.bb} (100%) delete mode 100644
> > > meta-isar/recipes-app/hello/hello_2.10.bb delete mode 100644
> > > meta-isar/recipes-app/hello/hello_2.9.bb
> > >
> > > --
> > > 2.17.1
> > >
> > > --
> > > You received this message because you are subscribed to the Google
> > > Groups "isar-users" group. To unsubscribe from this group and stop
> > > receiving emails from it, send an email to
> > > isar-users+unsubscribe@googlegroups.com. To view this discussion on
> > > the web visit
> > > https://groups.google.com/d/msgid/isar-users/20210331063235.6877-1-Vijaikumar_Kanagarajan%40mentor.com.
> > >
>
prev parent reply other threads:[~2021-03-31 18:53 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-03-31 6:32 Vijai Kumar K
2021-03-31 6:32 ` [PATCH v4 1/2] dpkg-base: Introduce do_apt_unpack Vijai Kumar K
2021-03-31 6:32 ` [PATCH v4 2/2] Modify hello to demonstrate apt:// without ${PV} Vijai Kumar K
2021-03-31 7:37 ` [PATCH v4 0/2] Custom ${S} series vijai kumar
2021-03-31 8:34 ` Henning Schild
2021-03-31 18:53 ` vijai kumar [this message]
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=CALLGG_LrSW9cE156Ee00nFmKbDY6WbC3kb6PBd4N43N2OqBPyQ@mail.gmail.com \
--to=vijaikumar.kanagarajan@gmail.com \
--cc=Vijaikumar_Kanagarajan@mentor.com \
--cc=henning.schild@siemens.com \
--cc=ibr@radix50.net \
--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