Re: Repo information inside rpm?

On Fri, 2004-08-13 at 10:55, Morten Kjeldgaard wrote:
> > Why do you want to extract an Apt location ?
> 
> That's a very good question! Well, first of all, I want to use it when I 
> build packages. When building, my packages end up in the place defined by 
> %_rpmdir and %_srcrpmdir. I am really tired of moving the packages by hand 
> to my apt directory. I keep making mistakes. If the package contains the 
> apt location, I can write a script to move it over.

In my case I have scripts that do the moving for me, so I don't think
this is a good reason for adding this particular format of the data to
the header. At least in my case I have enough information in the
filename itself to put things in the right place almost automatically.
The exception is which "repository" to move to (I have "planetccrma",
"planetcore" and "planetedge"). Syntax then becomes "bin/move
planetccrma". 

> Secondly, say you find a neat package somewhere, and say "I'd like more 
> stuff from this repo". Well, you can extract the apt location and see what 
> is there.
> 
> Thirdly, having the repo information there <em> will tell you where to go 
> to resolve dependencies </em>. This could be extremely useful in the 
> future as apt/yum tools become smarter and smarter.

That could be useful, but I would point to the "main" url of the
repository. 

Having the location of the package _in_ the package is a problem in the
sense that the package is no longer "relocatable" without rebuilding
(unless, as was mentioned in the thread, there is a tool that enables
you to change the headers of an already built package). And if a change
in url is needed, then all packages suddently contain stale information.
That is very bad.  

-- Fernando