Thanks Ted, committed as r150612 Are you OK with the cleanup to put the diagnostic header list & STR_SIZE helper macro in a common header (attached)? It'd be mildly annoying if (unlikely, I guess) we add a new header to one of these but not the other.
- David On Wed, Feb 15, 2012 at 11:08 AM, Ted Kremenek <[email protected]> wrote: > Looks great to me. > > On Feb 12, 2012, at 5:05 PM, David Blaikie <[email protected]> wrote: > >> On Thu, Feb 9, 2012 at 9:22 PM, David Blaikie <[email protected]> wrote: >>> On Thu, Feb 9, 2012 at 8:33 PM, Chris Lattner <[email protected]> wrote: >>>> On Feb 9, 2012, at 2:55 PM, David Blaikie wrote: >>>>> Here's a similar change to remove the diagnostic name from the main >>>>> table. It does mean removing a current clang flag, but I can't see a >>>>> particular use-case for this flag (& it was added in the same change >>>>> that originally added the two extra fields Benjamin just removed - so >>>>> I believe it's similarly unfinished work that never panned out) >>>>> >>>>> "Remove the unuseful -fdiagnostics-show-name >>>> >>>> I can't think of any useful purpose for this flag either. Does diagtool >>>> use it? >>> >>> Nope, though Diagtool does use the extra internal table >>> (StaticDiagNameIndex) of <name, diagid> - that was the other thing I >>> was going to fix (& I've done so) so here's a patch that includes both >>> the previous patch & the removal of StaticDiagNameIndex (or I should >>> say /moving/ since it now lives wholely in DiagTool that still needs >>> those name strings). There's a few things that could be tidied up >>> (this change means that ListWarnings.cpp in DiagTool has to do some of >>> the same work as DiagnosticIDs.cpp such as including all the >>> diagnostic tblgen'd headers (this could be added to a shared header) >>> and I copy/pasted the string size macro over there too (could be added >>> to the same header)) but it shows the principle >>> >>> It also brings Diagnostics.o to 430K (from 714K on my machine - not >>> the 697 you saw (on macos presumably, I'm on ubuntu Linux)) and brings >>> Clang's executable down by 0.5% (from 29.885M to 29.736M) >> >> Bump & now with documentation updates. >> >>> >>> - David >>> >>>> Ted, do you know if there are any clients of this? >>>> >>>> -Chris >>>> >>>>> >>>>> This option was added in r129614 and doesn't have any use case that I'm >>>>> aware >>>>> of. It's possible that external tools are using these names - and if >>>>> that's >>>>> the case we can certainly reassess the functionality, but for now it lets >>>>> us >>>>> shave out a few unneeded bits from clang. >>>>> >>>>> In a follow up change I'll see about removing the "StaticDiagNameIndex" >>>>> table >>>>> as well which should remove the actual diagnostic name strings from clang >>>>> entirely - though it'll mean moving that table into diagtool which still >>>>> needs >>>>> access to the string names to print its own results." >>>>> >>>>> On Thu, Feb 9, 2012 at 11:55 AM, Chris Lattner <[email protected]> wrote: >>>>>> >>>>>> On Feb 9, 2012, at 11:38 AM, Benjamin Kramer wrote: >>>>>> >>>>>>> Author: d0k >>>>>>> Date: Thu Feb 9 13:38:26 2012 >>>>>>> New Revision: 150199 >>>>>>> >>>>>>> URL: http://llvm.org/viewvc/llvm-project?rev=150199&view=rev >>>>>>> Log: >>>>>>> Kill the brief and full explanation fields from StaticDiagInfoRec. They >>>>>>> were unused and wasted space for nothing. >>>>>>> >>>>>>> - per PR11952. >>>>>> >>>>>> Nice, this shrinks DiagnosticIDs.o from 809K to 697K! >>>>>> >>>>>> -Chris >>>>>> _______________________________________________ >>>>>> cfe-commits mailing list >>>>>> [email protected] >>>>>> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits >>>>> <diag_opt_size.diff> >>>> >> <diag_size_opt.diff> >
diag_all.diff
Description: Binary data
_______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
