[76912] trunk/dports/devel/bzr-svn

[Ryan Schmidt]

On Mar 12, 2011, at 17:12, Rainer Müller wrote:

> On 2011-03-12 23:53 , Ryan Schmidt wrote:
> 
>> Also, consider that there are ports that set their version based on
>> other variables, for example based on ${svn.revision}. I don't
>> personally like it when ports do that -- I like the version field to
>> be the definitive source of the version, and to base any calculations
>> off of that, not the other way around -- but that's up to the
>> maintainers to decide for themselves. Having to shove calculations
>> like these into a line that already includes epoch and revision
>> information would probably make it less readable.
> 
> Splitting options is not the way we usually do it. For example, we set
> distname and extract.suffix which influence distfiles. We do not set
> distfiles and split distname and extract.suffix out afterwards.

One of our best-practices is to compute the branch out of the version field:

http://trac.macports.org/wiki/PortfileRecipes#branch


> Often a concatenation in the version field would be much easier to
> understand than splitting up version into svn.revision using nested
> split/lrange/join commands.

I suppose I am influenced by the desire to programmatically update a portfile. I have a script that does this, plugging a new version into the version line and new checksums into the checksum lines; the script fails when the version line isn't the real version but is a computation.