[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
proposed workaround (RE: Bidirectional repository synchronization with
proposed workaround (RE: Bidirectional repository synchronization with CVSup - how?)
Sun, 23 Sep 2001 08:40:57 -0500
Many thanks to Greg Woods and Eric Siegerman for setting me straight and
giving me several helpful suggestions.
Here is the approach and a procedure that I would like to take. (This is
not to say that Eric's suggestions were inferior, but it is simply more
tailored to our intranet environment.) Any comments on room for improvement
1. There exist several development sites (DS1, DS2, DS3, etc.).
2. Although only DS1 and the "grand master site" (GMS) are mentioned
here, the same pattern is repeated at other sites.
3. There are two master repositories (GMSL and DS1L), and there is one
replica (GMSR) maintained by CVSup.
4. GMSL is a master repository at GMS that is designated to hold
5. A copy of GMSL is maintained by CVSup at each development site.
While it will in fact have the same name in reality, herein I will
refer to it as GMSR to distinguish the copy at DS1 from GMSL at GMS.
6. DS1L is a "super-sandbox" repository at DS1, which is to say that it
is where untested changes may be committed. Its hierarchy
maintained identically with that of GMSL.
7. Having a single site for the CVS repository and using compression
do not give performance that is acceptable by all stakeholders.
That is why I am taking this approach.
Summary of the assumptions:
address@hidden <-- CVSup -- address@hidden
Procedure: Modify a single file that has no dependencies on other files.
1. Check file out from GMSR.
2. Update file with DS1L to ensure that no changes to this file have
been made in DS1L since the last time it was committed to GMSR.
3. Edit file*, make changes, and commit to DS1L.
4. Test file.
If fail, then go to previous step, else commit file to GMSL.
* Possibly, the developer may wish to issue the CVS edit command against the
file in both DS1L and GMSL, depending on how our procedure for using watches
- proposed workaround (RE: Bidirectional repository synchronization with CVSup - how?),
Art Eschenlauer <=