|
From: Ed M. <em...@fr...> - 2010-03-05 13:42:01
|
On Fri, Sep 11, 2009 at 10:06:38PM +1000, Nicholas Nethercote wrote: > On Fri, Sep 11, 2009 at 6:35 PM, Stanislav Sedov <st...@fr...> wrote: > > On Fri, 4 Sep 2009 22:22:25 +0400 > > Stanislav Sedov <stas@FreeBSD.org> mentioned: > > > >> Hi, guys! > >> > >> As you might know there has been work going on in the FreeBSD > >> community to port Valgrind to FreeBSD OS. ?Currently thanks to > >> efforts of a lot of developers we have a working preliminary > >> port done. ?Althought a lot of work has to be done yet, it is > >> in a good shape now and fairly usable on both amd64 and i386 > >> platforms. > >> > >> I'd love to work with someone from the community on integrating > >> our patches to the main tree to avoid excessive merge conflicts > >> in the future. ?There have been a lot of changes so that work > >> will require a good amount of time and coordination. > >> > >> Is it possible? ?How do we proceed? > >> > >> Let me know your thoughts. > >> > > > > So, is anyone interested? ?Where I should submit patches to? > > http://www.valgrind.org/info/platforms.html gives our general view on ports. > > In short, Valgrind ports are, as you say, a great deal of work. The > recent Darwin port represented years of work. And the audience for a > *BSD port is greatly smaller than Darwin. So the cost/benefit ratio > doesn't look very good. The patch is about 5800 lines of unified diff right now, and if I strip out build goo and trivial #ifdef change blocks about 4300 lines are left in the unidiff (~160 lines deleted and ~2200 added). The trivial #ifdef changes are cases where VGO_freebsd behaviour is identical to an existing case -- like the following: -#if defined(VGO_darwin) +#if defined(VGO_darwin) || defined(VGO_freebsd) I was actually surprised in going through this exercise that the change set is, relatively speaking, fairly small. The port's changes to Valgrind's internals are very limited. The port to FreeBSD is largely complete, relatively compact, and is not overly intrusive. There there are several active developers of the port. And there is certainly interest in the port - with 341 votes the bug containing the patch looks to be the highest-voted Valgrind issue. So I'd argue that the cost / benefit ratio is actually quite good. Regards, Ed Maste |