[PATCH 02/11] Support the new DiagnosticRemarks

Andreas Simbuerger simbuerg at fim.uni-passau.de
Tue Jun 10 04:14:01 PDT 2014


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

> 

>> Are you saying we call this even if -Rpass was not specified.
>> Did you check the compile-time impact of this on a larger file?
>> Or maybe the LLVM test-suite?
> 
> No, I have not tested the compile-time impact. Unfortunately we do
> not have access to the clang command line flags in opt afaik. I
> could check if the flag is enabled in opt easily but not in clang.
> 
> So either we emit them all the time (1), or emit them as soon as
> the user passes the polly-detect-track-failures option (2).
> 
> (1) might be bad in terms of compile-time. (2) is bad, if the clang
> user has to know that he needs polly-detect-track-failures in
> addition to -Rpass=".*"
> 
> Would be easier to have a way of checking that the remarks are 
> requested by the user.
> 
> 

Forget that, masking it with polly-detect-track-failures is perfectly
fine.

- -- 
Andreas Simb├╝rger
University of Passau - Programming Group
http://www.infosun.fim.uni-passau.de/cl/staff/simbuerger/

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (GNU/Linux)

iQEcBAEBAgAGBQJTluh5AAoJELS7a/6Ti2HYp0UH/RMReCdNy58NRpXW+TlylJTb
tAuKt+jvrlomUWo99s7i07H7zjQMGw1Ay0NQ3v1SdrxmpuVt4nwWVJOhuKfKdaCa
ZBPCevhhiD+QHioAn609DeGWEV52FPxa4a3b7sov+j/ke6sr/p2+xwdVYNSG05IQ
ALdpLBPstWGGxnOkjmxjWIIn4LKAlgetcRw/pNNoRWQf0CGqeSbqmR5CuS8lr287
B6/1VGQBe8iOu0eCwgs/SBnZlVgFysM4jMpOWYuABj60cZ/S0r68pYpTF5f3qAQ2
4L74aF1AUsFcpohEBfpL8sMyen4U5nC11nCDLznna9+6K0LizW1LqGYuErSgKhE=
=t3LT
-----END PGP SIGNATURE-----



More information about the llvm-commits mailing list