[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: RFC: quilt-el

* Mon 2006-07-03 Frank Küster <frank AT debian.org>
* Message-Id: 86bqs6h0k2.fsf AT alhambra.kuesterei.ch
> jari.aalto@cante.net (Jari Aalto+mail.linux) wrote:
>> I have found that the duo works best this way:
>> - init quilt
>> - make changes under quilt supervisions
>> - when satisfied, convert to dpatch 
>> [ repeat for as many patches as needed ]
>> I wouldn't use quilt alone, because it is stack based and I mostly try
>> to make all patches completely separate from each other.
> Well, that only works if you never patch the same file twice, doesn't
> it?  

It does. The order of patches in *.dpatch also stacks if needed. But
at all costs I try to avoid this, because it makes hard to drop
specific patch if upstream accepts the patched feature.

> Anyway, I would never use dpatch for my main packages,
> tetex/texlive.  It's just too slow to copy a 300 MB source tree (or will
> it be 600 when I'm working in a subversion checkout?)  

No need to copy ...

> or to diff
> between two of them just to add a one-line patch. 

You misunderstood. I use the duo: quilt + dpatch. The quilt for
intermediate editing, but the pieces are then converted in final *.deb
into *.dpatch format.

I have bash functions to automate all this, like picking up specific
quilt piece and then re-editing it and turning it into dpatch again.

In small projects, quilt is better than e.g. distributed VCS to handle
the source package (that is: working in separate branches).


Reply to: