Source: tsocks
Version: 1.8beta5-9.3
Severity: important
Control: clone -1 -2
Control: reassign -2 dpkg
Control: retitle -2 dpkg: should fail to create packages with invalid package 
relationships, e.g. Built-Using: libc6 (= )

Hi,

@tsocks:
tsocks builds packages on !linux architectures that are not accepted
into the archive since they contain invalid Built-Using entries like

  Built-Using: libc6 (= )


@deity:
dpkg should fail instead of creating packages that are not accepted
by the archive (unless --nocheck was given).


>From a discussion on debian-wb-team@:

On 2016-02-27 19:41, Adam D. Barratt wrote:
> On Sat, 2016-02-27 at 19:26 +0100, Samuel Thibault wrote:
>> Hello,
>>
>> Andreas Beckmann, on Sat 27 Feb 2016 15:58:29 +0100, wrote:
>>> could someone check the status of tsocks on hurd and kfreebsd, the
>>> packages are listed as "Uploaded" for "41 days". Something seems to need
>>> a little shaking to get them "Installed" :-)
>>>
>>> https://buildd.debian.org/status/package.php?p=tsocks&suite=unstable
>>
>> This is what I get in the REJECT ftpmaster mail:
>>
>>> tsocks-dbgsym_1.8beta5-9.3_hurd-i386.deb: APT could not parse Built-Using 
>>> field
>>
>> I don't know why tsocks is particular here.
> 
> According to the hurd-i386 log at
> https://buildd.debian.org/status/fetch.php?pkg=tsocks&arch=hurd-i386&ver=1.8beta5-9.3&stamp=1456586414
>  :
> 
>  Built-Using: libc6 (= )
> 
> The changelog says:
> 
>    * Document with Built-Using that tsocks ships the saveme binary,
>      that is statically linked against libc6 (Closes: #769343).
> 
> so I assume the patch is failing to take account of libc having
> different names across architectures. Indeed:
> 
> EXTRA_PACKAGES="libc6"
> echo -n "extra:Built-Using=" >> debian/tsocks.substvars
> for package in $EXTRA_PACKAGES; do
>   dpkg-query -f '${source:Package} (= ${source:Version}), ' -W $package 
> 2>/dev/null
> done >> debian/tsocks.substvars
> 
> CC-ing the NMUer, as this will need fixing in the package.
> 
> Regards,
> 
> Adam
> 

On 2016-02-27 22:12, intrigeri wrote:
> Adam D. Barratt wrote (27 Feb 2016 18:41:41 GMT) :
>> CC-ing the NMUer, as this will need fixing in the package.
> 
> Received, and read. I'm sorry for having messed up a NMU, and broken
> a couple interesting non-release archs.
> 
> Does this affect any release architecture?
> 
> If it doesn't, then given how my backlog looks like these days, I'm
> afraid I'll have to prioritize other matters. I'm not involved in
> porting to hurd and kfreebsd.
> 
> Cheers,
> 


Andreas

Reply via email to