11 messages in com.perforce.perforce-user[p4] build scripts| From | Sent On | Attachments |
|---|---|---|
| Usha Rajesh | 18 Jun 2001 16:34 | |
| Jeff A. Bowles | 18 Jun 2001 17:19 | |
| Rick Macdonald | 18 Jun 2001 20:09 | |
| Jeff A. Bowles | 18 Jun 2001 20:59 | |
| Chuck Karish | 18 Jun 2001 21:26 | |
| Paul C. Pharr | 18 Jun 2001 21:35 | |
| Dave Lewis | 19 Jun 2001 07:56 | |
| Rick Macdonald | 19 Jun 2001 08:19 | |
| Ines Heinz | 19 Jun 2001 08:54 | |
| Ines Heinz | 19 Jun 2001 09:00 | |
| Stephen Vance | 19 Jun 2001 19:59 |
| Subject: | [p4] build scripts |
|---|---|
| From: | Ines Heinz (hei...@llnl.gov) |
| Date: | 06/19/2001 08:54:51 AM |
| List: | com.perforce.perforce-user |
(Disclaimer: I'm not Jeff or Chuck. 8*) Chuck Karish wrote:
(Disclaimer: I'm not Jeff.)
(Disclaimer: I'm not Jeff or Chuck. 8*)
superstition, or mistrust of people, software and/or systems (build scripts, makefiles, dependencies, or something).
You bet.
Absolutely! I have had very good developers check in machine generated files. I have to work with a make system that doesn't clean all dependencies, leaves behind some template instantiations, doesn't remake a module if the makefile has changed unless you invoke the proper voodoo,...
I would agree that IF you could guarantee that the people, software, systems, and hardware were all working correctly, you shouldn't have to wipe out the nightly build directory and rebuild from scratch. However, I prefer to not presume that all of that will work correctly.
I'll turn this question around: Unless there are good reasons for not starting from scratch (a 10 hour build process for example), why not wipe out the directory?
Ines
Building in a fresh, clean client workspace helps ensure that all developers will be able to build in their own workspaces. Testing against such builds guarantees that the required behavior is actually implemented by changes stored in the depot.
I would add to this that there are possible changes in the hardware and system that are being tested as well.
Ines