Mike and Hyrum, On Mon, Apr 5, 2010 at 8:11 AM, C. Michael Pilato <cmpil...@collab.net> wrote: > Hyrum K. Wright wrote: >>> Thanks for the reminder. Merging the changed suggested here (r876245) does >>> indeed fix the test failure for me. >> >> Joe, >> I'm happy to +0 this patch, since it fixes the failure for me. I've no idea >> what it's doing under the hood, though, so I'd feel more comfortable if you >> were able to +1 it. > > Joe's changes are solid -- he's essentially just changed the Ruby wrapper > around txdelta transmission to ignore the digest. (It passes NULL to the > underlying C function so digest calculation never occurs.) And then he, of > course, no longer returns the uncalculated digest to callers. > > My only concern here is that it's a change to the Ruby API over what's > attempted in 1.6.x. If we're okay with that, you have my +1 to his fix. > But if we're trying to maintain compat in our bindings layers, we'll need to > do something different here. >
Mike, thanks for reminding me that I'd done this. I didn't like it at the time (when I thought it only impacted 1.7), but I'd forgotten all about it. I wasted a lot of time trying to make SWIG do it what I wanted, until I remembered the minimum of what really needed to be done. And though I suspect no one has been using this since it was terribly terribly broken (actually still is at least a little broken), you're right I shouldn't be changing interfaces. Hyrum, please merge r932942 on top of r876245. I think I've addressed Mike's concerns and I believe I haven't screwed up making stuff work for you. Since I've never been able to reproduce this particular problem, I'm uncomfortable proposing it for backporting until it makes it through a test run. If it works for you, let me know and I will propose the pair for backporting (or you may do it for me as I can't monitor the list that often). Thanks, -- Joe
