[Python-3000] not switching core VCS (was How to override io.BytesIO and io.StringIO with their optimized C version?)

Barry Warsaw barry at python.org
Fri Dec 28 19:07:25 CET 2007


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Dec 28, 2007, at 12:56 PM, Adam Olsen wrote:

> Providing official mirrors of all (or at least several important)  
> svn branches in popular DVCS tools would be sufficient.  Better, if  
> you consider that you can do several tools simultaneously, and  
> there's no transition necessary for the existing developers.

That's a transitional stage only, and in fact such beasts currently  
exist, so you could play with Python under <favorite dvcs> right now.   
But again, IMO, there are very big wins for even the official code  
base, and core developers, to want to host those branches natively  
under a dvcs.

Here's a simple one: how about not having your vcs force you to commit  
changes to a public repository before you're ready?  Under svn if you  
modify a file and before you commit it decide you want to rename it,  
svn puts up roadblocks, sometimes forcing you to commit your  
modification before you can do the rename.  bzr just does what you  
tell it to do.

> The word "distributed" in DVCS is exaggerated anyway.  You'll always  
> need a stable server to host public branches (as opposed to private  
> ones, which can just use your laptop's LAN ip address.)

It's distributed in the sense that connection to the public server is  
only necessary to publish change, not to make those changes, merge,  
branch, revert, commit, diff, etc. It's an important distinction  
worthy of the term, IMO.  Heck, that server isn't even necessary if  
you consider bundling branches (not patches!) into an email  
"publishing".

> The advantages related to merging are merely "nice" features, not  
> critical ones to warrant moving away from svn.

I respectfully disagree. :)

- -Barry

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (Darwin)

iQCVAwUBR3U7XXEjvBPtnXfVAQJnBAP/S9MJ3866WF2iSAS0pNrJBs102s3MqE1s
VP1ad3RvP/v9pGm9sG8IXjkl6ug2PV/gPsAXpkeq75ko7Dhn3toE+b0K8dctUOk3
RzkJ//cAUZwee/4smczKfET1Tu53UG4/gKNyzoPfSHpQHxQmDFZILxMy5o0dS3mm
erD+kGGwc94=
=UB+X
-----END PGP SIGNATURE-----


More information about the Python-3000 mailing list