From: Edward Diener (eldiener_at_[hidden])
Date: 2007-05-30 13:07:55
Rene Rivera wrote:
> Stjepan Rajko wrote:
>> I'm not an expert here but the following might be helpful:
>> The Boost sandbox is a subversion repository (see
>> http://subversion.tigris.org/), and is suitable for development and
>> version tracking of code. The vault, on the other hand, is a file
>> repository, and is suitable for posting of packaged versions of the
>> code (major milestones, finished products ready for review, etc) so
>> that they can more easily be used and reviewed.
>> Since the two serve very different purposes (and both are really
>> useful for that reason, IMO), I doubt they could be replaced by a
>> single area.
> Also, the vault is immediately open to all, whereas the sandbox requires
> permission from the moderators.
To me, since the idea is to put code into Subversion, which Boost
developers have now recognized as much superior to CVS, there should be
a single repository, or areas in a repository, whether called the
sandbox or vault, where new possible libraries for Boost may reside.
This would eliminate the confusion over two different areas of possible
future Boost code.
As far as submissions to that area is concerned, one could either give a
permission, ala the sandbox today, or allow anyone write access. One
could also have different policies in this regard for each different
area of a Subversion repository. On the read side, anyone running a
Subversion client would be able to access the code. Subversion clients
make it child's play to download what one wants from a particular
As far as there being any difference between version tracking and
packaging, Subversion can handle packaged code just like it can handle
version tracked code. There is nothing to keep someone from getting a
full zipped package from subversion if necessary.
My beef is that there are two totally different URLs and ways of access.
I think it would be better if there were a single one for future Boost
code, in whatever form it takes, and this could be accessed from a
Subversion client. Subversion is flexible enough to have packaged code
in an entirely different branch of a Subversion repository than
source/build code. But I would still like to see it all within the same
Once again, it would be much simpler if all future Boost code were in a
single place for easy access by interested people. It only confuses me
to hear that something is put into the vault, at some URL address I have
to remember and root around in, and other interesting libraries are put
into Subversion, at a particular repository. Sometimes KISS is wonderful
when it can be done easily, and Subversion is certainly up to that task.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk