[cfe-commits] r116312 - /cfe/trunk/test/Parser/MicrosoftExtensions.cpp
klimek at google.com
Sun Oct 17 06:39:20 PDT 2010
On Sat, Oct 16, 2010 at 11:30 PM, Chandler Carruth <chandlerc at google.com> wrote:
> On Wed, Oct 13, 2010 at 2:46 PM, Manuel Klimek <klimek at google.com> wrote:
>> Searching for svn:eol-style
>> "This means that by default, Subversion doesn't pay any attention to
>> the type of end-of-line (EOL) markers used in your files."
> Keep reading further in that section. My link is to the newer version of the
> book, but nothing has changed here:
> Basically, these properties are automatically set based on the file's
> deduced type.
I ran propget (for both mime type and eol-style) and proplist and
didn't see anything set for the file I tested. According to how I read
that section subversion only sets the mime type for non-textual
content by default.
>> A short check on a random file in clang:
>> svn-llvm/tools/clang/lib/AST$ svn proplist CXXABI.h
>> No output. If I don't completely misinterpret this, this means that
>> svn doesn't change EOL for those files. One solution would be to set
>> svn:eol-style to native on all code, which would fix the problem for
>> all clients.
>> On Tue, Oct 12, 2010 at 11:03 PM, Chandler Carruth <chandlerc at google.com>
>> > On Tue, Oct 12, 2010 at 10:31 AM, Francois Pichet <pichet2000 at gmail.com>
>> > wrote:
>> >> Is there any rule regarding new-line in LLVM?
>> >> I didn't see any in the coding standard and I notice that there is no
>> >> consistency at all across llvm/clang. Some files use \n while some use
>> >> \r\n
>> >> So i thought it doesn't matter.
>> > For text files, Subversion should be mapping line endings to the
>> > platform-native versions when syncing and committing from various
>> > platforms.
>> > I think one of the Windows clients is failing to properly do this, but I
>> > don't have a test or anything to demonstrate it.
>> > The result is that our (Google's) lit-like test runner gets confused by
>> > non-platform-correct line endings. We just fix them when we notice it.
>> > They
>> > should still show up as CRLF line endings on a windows machine when you
>> > run
>> > 'svn sync' (or equivalent for your client software).
>> > If I get some free time I make teach our test runner to not care, but
>> > ideally I'd fix the clients that aren't correctly handling eol-native
>> > SVN
>> > properties...
>> > _______________________________________________
>> > cfe-commits mailing list
>> > cfe-commits at cs.uiuc.edu
>> > http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
>> Manuel Klimek (http://go/klimek)
Manuel Klimek (http://go/klimek)
More information about the cfe-commits