On Tue, 2007-09-18 at 22:36 -0400, Mike Frysinger wrote:
> On Tuesday 18 September 2007, Ned Ludd wrote:
> > qtbz2 -O -x /tmp/wget-1.10.2.tbz2 | qxpak -l -O -
> 
> one thing ive never liked about qtbz2 and qxpak is the usage seems obscure 
> ... 
> i need to re-learn the syntax everytime i try playing with them

likewise.

> can it be improved ?  ive thought about it but havent figured out a good 
> solution ... maybe just add copy & paste examples to the manpage ?  or change 
> the syntax of qxpak to mimic tar ?  or something else ?

I'd rather us avoid changing syntax as that can break scripts. Overall 
the tools work really well after you master the syntax down for the 
operations you need them to preform.

What kills me the most is extracting a single xpak item to stdout. 
In order do that we have do something like.

qtbz2 -x -O wget-1.10.2.tbz2 | qxpak -O - -x CFLAGS

But in reality it's something more like

for x in $(qtbz2 -x -O wget-1.10.2.tbz2 | qxpak -O -l - | grep -v
-e .ebuild$ -e .bz2$) ; do echo $x=\"$(qtbz2 -x -O wget-1.10.2.tbz2 |
qxpak -O - -x $x)\" ; done

And that is a crap load of execve() and pipes just to list fields 
for one tbz2 file. For the tbox's html output we do that more or 
less for about 10k binpkgs

Maybe the -x option should take ITEM1,ITEM2,ITEM3.. and process them in
order. 

Granted right now we can do
qtbz2 -x -O wget-1.10.2.tbz2 \
  | qxpak -O - -x RDEPEND CFLAGS CXXFLAGS USE DEPEND

But the output ordering is not the same as we requested on cmdline. It 
appears to be based on the ordering that the item shows up in the vdb. 

Shrug sorry I don't think I really answered the question but this 
is all that came to mind..

-- 
Ned Ludd <[EMAIL PROTECTED]>

-- 
[EMAIL PROTECTED] mailing list

Reply via email to