The current status is that this is too slow, so for now the recommendation
is to have a case insensitive mount for the files that need it.
On Sep 9, 2016 8:18 PM, "John Sheu" wrote:
> sheu added a subscriber: sheu.
> sheu added a comment.
>
> I'd be very interested in seeing this patch happen.
sheu added a subscriber: sheu.
sheu added a comment.
I'd be very interested in seeing this patch happen. What's the current status
here?
Also, w.r.t. the cache invalidation problem -- would it be feasible / a good
idea to move users of the FileSystem API to VirtualFileSystem, in general?
Thi
On Fri, Jun 10, 2016 at 10:35:26AM -0700, Hans Wennborg wrote:
> On Thu, Jun 9, 2016 at 5:31 PM, Joerg Sonnenberger via cfe-commits
> wrote:
> > On Wed, Jun 08, 2016 at 01:12:31AM +, Hans Wennborg via cfe-commits
> > wrote:
> >> There has been a patch for this before (http://reviews.llvm.org/
On Thu, Jun 9, 2016 at 5:31 PM, Joerg Sonnenberger via cfe-commits
wrote:
> On Wed, Jun 08, 2016 at 01:12:31AM +, Hans Wennborg via cfe-commits wrote:
>> There has been a patch for this before (http://reviews.llvm.org/D2972),
>> but this one is more general, as it works on the virtual filesyst
On Wed, Jun 08, 2016 at 01:12:31AM +, Hans Wennborg via cfe-commits wrote:
> There has been a patch for this before (http://reviews.llvm.org/D2972),
> but this one is more general, as it works on the virtual filesystem
> layer, and also supports case-insensitive lookups of parent directories
>
hans updated this revision to Diff 60220.
hans added a comment.
Adding the version that caches directory contents.
This has the problem the it doesn't play with modules, the rewriter, and
possibly others, because they write to the file system without any way for the
vfs to find out. I've tried
hans added a comment.
In http://reviews.llvm.org/D21113#452673, @thakis wrote:
> Not sure if we want a flag that adds 50% overhead, no matter how convenient
> it might be :-/
I now have a version of the patch that caches directory contents that it lists:
real2m31.461s
user68m42.09
thakis added a comment.
Not sure if we want a flag that adds 50% overhead, no matter how convenient it
might be :-/
http://reviews.llvm.org/D21113
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinf
hans added a comment.
In http://reviews.llvm.org/D21113#452558, @thakis wrote:
> Can you try building a few more files? Say, v8_base?
Well, that was depressing:
Putting the sdk on a vfat fs:
real2m26.077s
user68m31.476s
sys 1m25.702s
Using the flag:
real9m5.179s
thakis added a comment.
Can you try building a few more files? Say, v8_base?
http://reviews.llvm.org/D21113
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
hans added a comment.
In http://reviews.llvm.org/D21113#452431, @thakis wrote:
> For performance: Can you check how build times for some large target in
> Chromium on Linux targeting Windows compares with this vs having the sdk in a
> fat mount? That would give us some data.
I measured compil
thakis added a comment.
For performance: Can you check how build times for some large target in
Chromium on Linux targeting Windows compares with this vs having the sdk in a
fat mount? That would give us some data.
The discussion in the include case warning thread sounds like MS might update
i
hans updated this revision to Diff 60054.
hans marked 2 inline comments as done.
hans added a comment.
Addressing Saleem's comments, and renaming the flag to
-fcase-insensitive-paths, since this doesn't apply just to includes, but
anything that goes through vfs, including the main source filenam
compnerd added a comment.
This looks good to me. Thanks for picking this up! And thanks for the perf
numbers!
Comment at: lib/Basic/VirtualFileSystem.cpp:401
@@ +400,3 @@
+IntrusiveRefCntPtr Base)
+: Base(Base) {}
+
Probably can inline this in the hea
hans updated this revision to Diff 59982.
hans added a comment.
Add --show-includes test.
http://reviews.llvm.org/D21113
Files:
include/clang/Basic/VirtualFileSystem.h
include/clang/Driver/Options.td
include/clang/Lex/HeaderSearchOptions.h
lib/Basic/VirtualFileSystem.cpp
lib/Driver/To
15 matches
Mail list logo
