Hi Alan,
I don't see that there is any reason or need for the code base to fork.
My preference is for both repositories to:
- pull regularly from each other
- minimize differences
- rapidly evolve helpful new features
I hold two positions regarding SGI in this matter:
1) In respect of SGI's contribution, their repository _is_ and _should_
be canonical.
2) The user community cannot expect SGI to additionally bear the expense
of fully supporting - current, packaged, tested - versions of PCP for
all other manufacturers' systems.
In gratitude for SGI's initial contribution, the user community must be
mindful of enhancing PCP in a way that is helpful to SGI, and acceptable
by them. The user community benefits by having easily deployable PCP
packages for many different platforms; this is the reason behind
toolworks@sourceforge. Through SourceForge, diverse hardware resources
are available, and there is sufficient access to facilitate a substantial
volume of concurrent contribution from various domain specialists.
Those specialists should be able and attentive to 1) making changes that
are palatable to SGI, and 2) successfully negotiating the differences.
In this way, the user community can give back to SGI, who has given
so much to them.
Regarding the recent SourceForge changes: I'm happy to send Mark a diff,
if he prefers. I surmise that he prefers to simply take the SourceForge
changes at his convenience, by directly diff'ing the tags in the
SourceForge repository. All the necessary information is available on
the SourceForge sites. I notified Mark directly ten days before the
announcement; he has made no additional request or inquiry. I have
endeavored to provide what he needs to do his job easily. If he wants
anything additional, of course I will help him - this is the purpose of
this project.
Do you have any more questions or concerns? Is there anything you think
should be done differently? Let me know.
- mtw
On Fri, 6 Jan 2006, Alan Hoyt wrote:
Michael,
Could you please clarify – are you forking the code base or were you going to
provide a diff patch to Mark so he can fold these changes into the SGI code
line?
Alan
Michael Werner wrote:
This message is to announce the availability of prepackaged binaries
of PCP 2.5.0 and PCPMon 1.3.4 for Fedora 4 and MacOS 10.3 and 10.4
The packages, sources, docs, etc. are available from:
http://reliable.sourceforge.net
http://sourceforge.net/projects/reliable
Packages for additional platforms are in the works. Your early
questions, comments, and feedback will be most helpful and welcome.
This PCP 2.5.0 distribution contains modifications not found in SGI's
original PCP 2.5.0 sources. Source code and details of the individual
changes can be obtained by visiting:
http://reliable.sourceforge.net/pcp/build_info/pcp_2_5_0.html
Generally speaking, the differences are:
* Fixed MacOS X compilation errors
* Fixed MacOS X PMDA network routines to also work on MacOS 10.4
* Added files and scripts for packaging and installation on MacOS X
* Added and lightly refurbished Michal Kara's MySQL PMDA, previously
moribund, so PCP can now provide a LAMP monitoring solution.
This is the first release of PCPMon since version 1.3.3 was released
by Michal Kara in 2001. Source Code and details of the individual
changes can be obtained by visiting:
http://reliable.sourceforge.net/pcp/build_info/pcpmon_1_3_4.html
Generally speaking the differences are:
* Ported to MacOS X
* Added help menu items linking to additional web documentation
* Fixed autotools build process
* Fixed several runtime hangs and crashes
* Added several loadable config files for easily displaying
common metric graphs
Cheers,
- mtw
|