Bug#1026199: release.debian.org: Is the toolchain list updated for bookworm

[Guillem Jover <guillem@hadrons.org>]

Hi!

On Sat, 2022-12-17 at 08:42:18 +0100, Paul Gevers wrote:
> On 16-12-2022 02:40, Sam Hartman wrote:
> > I was looking at https://release.debian.org/testing/essential-and-build-essential.txt
> > 
> > trying to figure out which packages I'm involved in are covered by the
> > toolchain freeze.  I am wondering what's still pulling
> > libgssapi-krb5-2 and friends into build-essential.  It used to be
> > pulled in via pam via libtirpc, but that should have gone away with
> > the pam upload of 1.4.0-13.
> > 
> > 
> > I'm wondering if that list hasn't been recently updated or if there's some other dependency cycle pulling in krb5?

Ah, was wondering the same few days before the toolchain freeze, as I
was unsure whether to update some of the packages I maintain (in
particular libmd, for which I was thinking of doing a new upstream
release), and was also checking hints for explicit blocks. See below…

> I just refreshed the list, it's still there. I used a script on udd,
> essentially this:
> 
> essentials = {'build-essential'}
> 
> def add_sql_packages(con, essentials, query):
>     cur = con.cursor()
> 
>     cur.execute(query)
>     items = cur.fetchall()
>     cur.close()
> 
>     for item in items:
>         if item[0] is not None:
>             no_alt = re.sub(alternatives, "", item[0])
>             no_ver = re.sub(version, "", no_alt)
> 
>             for x in no_ver.split(','):
>                 essentials.add(re.sub(multiarch, "", x.strip()))
> 
> query = "SELECT DISTINCT package FROM packages WHERE release = 'bookworm'
> and essential = 'yes'"
> 
> add_sql_packages(con, essentials, query)
> 
> done = False
> while not done:
>     ess_org = copy.copy(essentials)
>     query = """SELECT DISTINCT depends FROM packages WHERE release =
> 'bookworm' and
> package in ('%s')""" % "','".join(essentials)
>     add_sql_packages(con, essentials, query)
>     if essentials == ess_org:
>         done = True

…but hmm, is this perhaps not taking into account Pre-Depends?

Thanks,
Guillem