llvm dev - Aug 2014 - [LLVMdev] sys::path::system_temp_directory vs. sys::fs::createTemporaryFile

On 25 July 2014 17:57, David Majnemer <david.majnemer at gmail.com>
wrote:
> To be clear, I don't think we need both; I'd like for us to drop
one of
> them.  I just want to make sure that that the final state, post merge,
works
> well on all of our platforms. :)
Cool. The attached patch removes TempDir. It passes all tests, but the
OS X case looks a bit suspicious. I first implemented it by given
priority to confstr, but we depend on the env variables taking
priority for testing the crash report. Is there some way to change
what confstr returns?

Should we just add a -cc1 command line option for use in
test/Driver/crash-report.c?

Cheers,
Rafael
-------------- next part --------------
A non-text attachment was scrubbed...
Name: t.patch
Type: application/octet-stream
Size: 7482 bytes
Desc: not available
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20140808/344b4fe5/attachment.obj>

On Aug 8, 2014, at 18:35 , Rafael Espíndola <rafael.espindola at
gmail.com> wrote:
> On 25 July 2014 17:57, David Majnemer <david.majnemer at gmail.com>
wrote:
>> To be clear, I don't think we need both; I'd like for us to
drop one of
>> them.  I just want to make sure that that the final state, post merge,
works
>> well on all of our platforms. :)
> 
> Cool. The attached patch removes TempDir. It passes all tests, but the
> OS X case looks a bit suspicious. I first implemented it by given
> priority to confstr, but we depend on the env variables taking
> priority for testing the crash report. Is there some way to change
> what confstr returns?
> 
> Should we just add a -cc1 command line option for use in
> test/Driver/crash-report.c?
Thanks for doing this, Rafael!

It looks like OS X always has TMPDIR set these days (to the value of
_CS_DARWIN_USER_TEMP_DIR), so I think we should just go with TMPDIR if it's
set and non-empty, and then fall back to confstr. However, we should not be
using TMPDIR if ErasedOnReboot is false (i.e. if we'd prefer
_CS_DARWIN_USER_CACHE_DIR). Unfortunately there's no canonical environment
variable for that.

CCing Greg to see if that makes sense.
Jordan
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20140812/3c585a91/attachment.html>

> It looks like OS X always has TMPDIR set these days (to the value of
> _CS_DARWIN_USER_TEMP_DIR), so I think we should just go with TMPDIR if
it's
> set and non-empty, and then fall back to confstr. However, we should not be
> using TMPDIR if ErasedOnReboot is false (i.e. if we'd prefer
> _CS_DARWIN_USER_CACHE_DIR). Unfortunately there's no canonical
environment
> variable for that.
>
> CCing Greg to see if that makes sense.
OK, so I guess something like this?

Cheers,
Rafael
-------------- next part --------------
A non-text attachment was scrubbed...
Name: t.patch
Type: text/x-patch
Size: 7578 bytes
Desc: not available
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20140813/1a3248cf/attachment.bin>