Evgeni Golov wrote:
> Hi,
>
> [snip]
>
>> The problem being that upstream author, Marco D'Itri, does not agree
>> with bindgraph being packaged for Debian.
>> Some time ago, I had to repackage bindgraph source (that's where the 'a'
>> suffix comes from) to include the changes with respect to the original
>> ---which is unmodified since AFAIK ---
>> The patches are there to document the deviations from "pure upstream".
>>
>> ... which might mean a need for a README.Source, I agree
>>
>> Only options are, hence:
>> - leave them as documentation of deviations from upstream ("less bad")
>> - revert to upstream source 0.2 + patches ---> absurdly big debdiff
>> for a trivial change, plus versioning paradox
>> - publish a new "completely unofficial" 0.3 version, including all
>> changes and/or upstream + patches applied at build-time
>>
>> but I might be missing some other option. Any suggestions are welcome,
>> of course
>>
>
> Well, I think option 2 isn't that bad when using format 3.0 ;)
> But on the other hand, you have it working like this now, and one should
> never change a running system.
>
In general, unless I have a "regular" sponsor, I tend to minimize
debdiff changes to make it easier for a sponsor to review the changes.
I can of course prepare a version where we have the original 0.2 plus
all changes as patches, but that would be quite paradoxical w.r.t. naming.
However, "0.2+a" might be a sane name, and indeed would sort later than
"0.2a". I can prepare a version from that, and resubmit including the
diff against vanilla 0.2.
(in this case, source format 3 more than justifies itself :-D )
> Thats also why I don't understand why you convert to 3.0 now, you don't
> have any patches to apply besides debian/, and here you do not gain
> anything over the old version. Thus I'd stay with 1.0, fix the stuff you
> want to fix, and upload that one then. But it's your package, so you are
> free to decide to go to 3.0.
>
It was a huge gain for couriergraph, and I figured it would be best for
bindgraph.
In any case, I'm trying to convert all packages where it doesn't give
problems to the new format: it is indeed the best way to document
deviations from upstream, plus the added bonus of not having to
repackage upstream bzipped tarballs.
>> If you sponsored couriergraph too, I'd be more than glad, too ;)
>>
>
> Not today, and most probably not tomorrow, but I can have a look on that
> if you don't find anyone in the meantime ;)
>
Thanks in advance.
--
To UNSUBSCRIBE, email to [email protected]
with a subject of "unsubscribe". Trouble? Contact [email protected]
