public inbox for isar-users@googlegroups.com
 help / color / mirror / Atom feed
From: Henning Schild <henning.schild@siemens.com>
To: "[ext] Jan Kiszka" <jan.kiszka@siemens.com>
Cc: <isar-users@googlegroups.com>
Subject: Re: [RFC Kernel PATCH 1/1] builddeb: support creation of linux-perf packages
Date: Wed, 4 Sep 2019 10:37:01 +0200	[thread overview]
Message-ID: <20190904103701.0f78b888@md1za8fc.ad001.siemens.net> (raw)
In-Reply-To: <acd80d03-73b0-a8fa-f4cb-5753c3898817@siemens.com>

Am Tue, 3 Sep 2019 17:19:14 +0200
schrieb "[ext] Jan Kiszka" <jan.kiszka@siemens.com>:

> On 03.09.19 16:59, Baurzhan Ismagulov wrote:
> > Hello Cedric,
> > 
> > On Tue, Sep 03, 2019 at 10:01:43AM +0200, Cedric Hombourger wrote:  
> >> Unfortunately Ubuntu does not use the same package structure  
> > ...  
> >> Anything comes to mind?
> >>
> >> Should we give up the idea of using builddeb and instead use our
> >> own debian/{control,rules} in Isar? (Jan told me that you have
> >> started to discuss/consider this)  
> > 
> > Thanks for summarizing the differences. I'm afraid there is no
> > silver bullet, it will diverge. As I see it, the approach matrix is
> > still the same:
> > 
> >            generic  specific
> > 
> > own       complex  simple
> >                     repetitive
> > 
> > upstream  complex  complex
> >            fragile  fragile
> > 
> > For a single downstream project, I'd choose own-specific at the
> > cost of some repetition. I assume it to require less effort than
> > upstream-generic. In my understanding, this is what Ubuntu does due
> > to practical reasons. Unifying the three upstreams is a noble task,
> > but I don't see anyone feeling inclined to solve that for the
> > long-term.
> > 
> > For Isar or a generic base layer on top of it, the effort mix will
> > be different. Upstream-specific doesn't make sense, and
> > upstream-generic is difficult due to the reasons in your mail. The
> > alternative is copying own-specific. Given enough permutations,
> > this will become unmanageable at some point (updating will be
> > costly and error-prone). Start with own-generic via e.g. sed
> > <config.in >config -- and you end up with your own version of
> > deb-pkg. So, here the question is how many variants one is going to
> > ultimately have.
> > 
> > For the generic case, I don't see approaches better than deb-pkg
> > ATM. IMHO, it doesn't have to be ugly. They already support e.g.
> > rpm; do they have variants for Fedora, SUSE, etc.?
> >   
> 
> Addressing also Ubuntu would by a nice by-catch but not a must-have.
> We need to get Debian right first.

I do not care about ubuntu either, but i reviewed a kernel patch where
not wearing my isar-glasses.

Having support for building perf as a debian package is a nice upstream
feature anyways, no matter if isar uses it. Or keeps using it in the
future.

> Yes, moving away from builddeb is seriously considered as the current
> model makes the kernel build also "one off" /wrt all other packages.
> Plus we repackage anyway because upstream is not generic enough to
> allow us modeling the packages in a way that makes them drop-in
> replacements for the Debian kernel. And having the control downstream
> would also make it easier to account for small differences in the
> various Debian flavors - if there is a user.

Agreed. That repackaging and other custom stuff makes clear that
upstream kernel does not fit our needs. And we can not get our patches
for some aspects upstream.

Henning

> But creating our own kernel recipe is not highest prio ATM, there is
> bigger fish to fry first.
> 
> Jan
> 


      reply	other threads:[~2019-09-04  8:37 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-08-28  7:17 [RFC Kernel PATCH 0/1] " Cedric Hombourger
2019-08-28  7:17 ` [RFC Kernel PATCH 1/1] builddeb: " Cedric Hombourger
2019-08-30 13:58   ` Henning Schild
2019-09-03  8:01     ` Cedric Hombourger
2019-09-03 14:59       ` Baurzhan Ismagulov
2019-09-03 15:19         ` Jan Kiszka
2019-09-04  8:37           ` Henning Schild [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=20190904103701.0f78b888@md1za8fc.ad001.siemens.net \
    --to=henning.schild@siemens.com \
    --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