Hey :), On mié, 2012-02-01 at 14:45 -0500, Ehsan Akhgari wrote: > On Wed, Feb 1, 2012 at 1:58 PM, Jean-Marc Desperrier <[email protected]>wrote: > > > Hi, > > > > At the end of last year, there was some indexing problems on the dxr on > > Lanedo. After checking again after Taras's last blog entry, I see now it > > seems to work much better : > > http://dxr2.lanedo.com/search.**cgi?tree=mozilla-central&** > > string=GfxContext<http://dxr2.lanedo.com/search.cgi?tree=mozilla-central&string=GfxContext> > > http://dxr.mozilla.org/**mozilla/search.cgi?tree=**mozilla-central&string= > > **GfxContext<http://dxr.mozilla.org/mozilla/search.cgi?tree=mozilla-central&string=GfxContext> > > > > What solved it exactly ? Christian Dywan did refer to it being broken > > because of the use of the latest llvm and clang, but Ehsan replied it was > > working correctly with clang 3.0. > > So it's not very clear which version should be avoided. > > > > clang 3.0 should be fine.
We have stick so far to r132484 for clang, which isn't a final 3.0 revision. the problems Christian mentioned involved using more recent clang/llvm, where API changes broke the clang plugin build. > > > > Another thing, a small bit of GUI nitpicking : the type of functions and > > variables was added, which is good, but as a result it's more frequent for > > the value to be split on 2 lines, and it's not always obvious it's part of > > the same line. For example when seeing : > > "class gfxContext * > > nsSVGRenderState::**GetGfxContext" > > it's not obvious it's only one line that was split in two. > > But anyway there's obviously some adjustments to do to handle overflow > > better in that part. > > > > Please file a bug on this, it looks like a UI issue. Looks like an unintentional behavioral change in dxr.lanedo.com, I'll look into that. Regards, Carlos _______________________________________________ dev-static-analysis mailing list [email protected] https://lists.mozilla.org/listinfo/dev-static-analysis
