[p4] handling different lineEnds - Richard Ramos - com.perforce.perforce-user

11 messages in com.perforce.perforce-user[p4] handling different lineEnds

From	Sent On	Attachments
Richard Ramos	22 Oct 2002 19:19
Chuck Soper	22 Oct 2002 20:56
Oren Shemesh	22 Oct 2002 23:01
Richard Ramos	23 Oct 2002 06:48
Dave Lewis	23 Oct 2002 08:31
Chuck Karish	23 Oct 2002 09:35
Dan Kegel	23 Oct 2002 10:22
Chuck Soper	23 Oct 2002 12:49
ste...@vance.com	23 Oct 2002 17:13
Chuck Karish	23 Oct 2002 21:59
Chuck Soper	24 Oct 2002 18:01

Subject:	[p4] handling different lineEnds
From:	Richard Ramos (rra...@elitecorpusa.com)
Date:	10/23/2002 06:48:19 AM
List:	com.perforce.perforce-user

Thanks. This is exactly what I was looking for. I had checked out the file and changed the type to binary but discovered afterward that I couldn't diff. That would have been a problem.

As for Tornado, it's quite a mess. The error messages when it encountered a prj file with CRLFs was completely unhelpful. I ended up running the wtxtcl interactive shell and manually stepping through tcl scripts before figuring out that it just wasn't reading the files correctly.

Thanks again.

-----Original Message----- From: perf...@perforce.com [mailto:perforce-user-admin at perforce.com]On Behalf Of Oren Shemesh Sent: Wednesday, October 23, 2002 2:02 AM To: perforce-user at perforce.com Subject: RE: [p4] handling different lineEnds

Hi Richard,

We had the same problem here.

Our solution was declaring all such files as binary. Here are two lines from the typemap table:

binary+wl //Depot/....wpj binary+wl //Depot/....wsp

Of course, you must add those files as binary in the first place, changing them from text to binary after they are checked-in is no good. (If you already have them, then change their type to binary, then over-write them with a good, LF-only copy of the files).

You can solve the problem of diffing those files by going to the Options dialog, selecting the 'Diff/Merge' tab, and cheking the 'Allow diff of binary files' box :-). (It exists on P4Win build 35186 - may be you do not have it if you are running an older build). It allows comparing binary files both with the built-in P4Diff, and with external utilities.

I think that your real problem would be merging those files, assuming that multiple developers modify the same revision. As you could see, we prevent that by declaring them +wl. after realizing that Tornado is so sensitive to the line-ends, we decided never to try and merge those files.

Feel free to bug me - I know I would be glad to find someone who works with both Perforce and VxWorks. It would have saved me a lot of time debugging this problem. An anecdote: After getting the faulty .wpj files from Perforce (With CR/LF) for the first time, the phenomenon we saw was the linker consuming all available memory and forcing us to reboot the PC. It took more than a few hours to understand that the problem was the #*$&&% .wpj files with CR/LF. It got me angry as Tornado enough to develop a build system of my own (Based on GNU Make, using Cygwin), which we now use happily. Since all our project files are simple text file, there is no problem merging them, and so people are free to work as they wish, no more racing to change the .wpj files :-)

Oren.

-----Original Message----- From: Richard Ramos [mailto:rramos at elitecorpusa.com] Sent: Wednesday, October 23, 2002 5:20 AM To: perforce-user at perforce.com Subject: [p4] handling different lineEnds

I'm having a problem with the way perforce handles end of lines in a mixed file (dos and unix)environment.

I'm working on a system built on vxWorks that uses tornado project files. Apparently, Tornado does not like to see CRLFs in their project files. It chokes unless I set the file format to unix. Unfortunately, since the project files are text files, as are my source files, their lineEnd treatment is uniformly determined by a client attribute which I have set to "local". When I check the files out they end up with CRLFs.

I thought a workaround would be to change the project file types to binary but then I can't diff the files using the "diff file against depot" command. Can someone suggest a simple solution to this problem?

Thanks.

Rich Ramos