Den 25. feb. 2009 01.17 skrev Roy Bynum <rabynum@ieee.org>:

David,

I am sure that everyone in the group appreciated your and others efforts to support Mono applications.  I am interested in your comment about shaving off some of the size of the Mono "stack".  We have been having the discussion about perceived desktop performance.  In years past, as a real time systems programmer, I became very aware that one of the primary performance issues is the size of the code that is being executed.  In reducing the size of the "stack", will you also be providing something of an inprovement to the perceived performance of the applications using that "stack"?

The biggest win I suspect will be fixing the debug stripper and then introducing a mass rebuild of the mono packages. I unfortunately am not qualified to perform that change (I believe a bug exists for this). Then we might be able to gain some savings by looking hard at our support, I haven't looked into this suffiencently to see if it applies to the Fedora Mono stack but by cutting out the obsoleted Mono 1 support from the builds we should be able to cut some 15% from the existing packages. This last one comes from brief conversations with directhex who does Mono work in Ubuntu and has helped coordinate similar work between Ubuntu and Debian, I have not yet been able to get ahold of our Mono maintainer to determine if this is suitable for us.

Aside that, I have noticed that as versions get bumped and libraries are replaced the spec files aren't getting updated correctly always. Having a review of what we truly need to depend on is valuable. I was able to cut a bit of fat from the banshee build already this way.

I am hoping an approach of little gains here and there will both improve the Mono stack and get us closer to reaching the cd size limit. At any rate it is valuable work.

Also on the list of things to do currently is fixing up the Mono packages to actually ship their documentation and update monodocs index, these will become separate packages and should not add any fat though.

All of this though should be fairly irrelvant when it come to the debate at hand, namely the technical merits of Banshee. I have contacted the two main upstream maintainers and they have both kindly agreed to attend an eventual IRC meeting to help answer questions as well as expanding the feature proposal page.

- David