User Details
- User Since
- Mar 30 2017, 7:33 AM (338 w, 5 d)
Jan 8 2019
We have both explicitly and implicitly prebuilt modules, but this patch only relates to implicit ones. I agree that our usecase is a bit unique where we're installing and distributing implicitly built pcm, but I also imagine there might be a usecase where you built implicit pcms in some directory and want to reuse them somehow from other locations. If it doesn't make sense to be upstream, this can stay in our fork.
Jan 7 2019
Nov 5 2018
Oct 24 2018
Landed in rL345121
Updated HasSpace, added comments, fixed typo in the commit message
Oct 23 2018
Sep 25 2018
Aug 1 2018
Jul 26 2018
Jul 10 2018
Jun 26 2018
Delete a.out)
Add #import test and add branch AllowExtraModuleMapSearch when ModuleName was from @import
Jun 25 2018
use %clang_cc1 instead of %clang
Jun 21 2018
Add a test case
Jun 20 2018
Update commmit message and comment, add ObjC2
May 23 2018
Apr 26 2018
Apr 25 2018
Do you mean something like addDeserializationListener which create a multiplex listener with an existing listener? I think that'll be good as well!
Apr 21 2018
Mar 16 2018
@marsupial Thanks for the commment, found a bug in my script and fixed. Now this patch removes only includes which are not used in each TU.
Mar 15 2018
Mar 7 2018
Fixed typos
Mar 5 2018
Nov 19 2017
LGTM, thanks!
Nov 3 2017
Made a trivial fix in the testcase
Update diff. Pass all flags just with "," including values.
Oct 27 2017
Oct 26 2017
Sep 5 2017
LGTM! Thanks for the fix!
Aug 29 2017
Aug 28 2017
Update diff. I agree that this is more generic.
D37217 has already landed, so we don't need this anymore.
I think this way is more reasonable than mine. Thank you very much!
Aug 23 2017
Aug 18 2017
Update diff. Thank you for your comments!
Aug 17 2017
Update diff.
Update diff.
const char* Values -> const char *Values
Update diff.
Aug 16 2017
@ruiu
I understand your concern. However by doing this (rather than building functions for each flag), we will be able to get all information in OptTable itself. It is true that this is quite fragile, but adding ValuesCode to new flag is not complicated, so I believe overall it's fine.
Update diff.
Update diff according to Raphael's comments.
Aug 15 2017
Aug 11 2017
Aug 9 2017
Aug 2 2017
Fixed typo.
Aug 1 2017
Modified comment.
@ruiu
This test will break for instance, when someone add new value to mrelocation-model because values has to be shown in row 78- 83 order, or when someone made a new flag which start with -Wno-invalid-pp- because in row 96 only -Wno-invalid-pp-token is expected for this prefix.
Update diff.