Andi Vajda wrote:
> On Thu, 9 Nov 2006, Heikki Toivonen wrote:
> I know, I know.... the logic is incomplete though, if svn tends to be
> unavailable more often than builds, then putting all eggs into the svn
> basket
> would be unwise. I sure don't know that that is the case, I was just not
> convinced by the argument...

I don't understand that logic.

Currently:

svn | builds | can checkout and build
=====================================
no  | no     | no
no  | yes    | no
yes | no     | no
yes | yes    | yes

All in svn:

svn | builds | can checkout and build
=====================================
no  | no     | no
no  | yes    | no
yes | no     | yes
yes | yes    | yes

>> Like I mentioned, when you use svn+ssh. By its nature it encrypts the
>> connection and verifies that nothing changes over the wire.
> 
> How is that more secure than downloading source tarballs over ssh, https ?

It isn't, but we are downloading the tarballs over http, not ssh or
https. We could probably make downloads reasonably easy over https
(would need to enable SSL on builds, checkin a cert to chandler/ and
make curl use that cert when doing the SSL checks.)

> My main concern is that I'm not sure svn is such a great tool for
> handling large files. I could be wrong... If we follow this route,

I believe my tests showed that download speeds should not be an issue.
If you want me to run other tests I'd be happy to.

-- 
  Heikki Toivonen


Attachment: signature.asc
Description: OpenPGP digital signature

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

Open Source Applications Foundation "chandler-dev" mailing list
http://lists.osafoundation.org/mailman/listinfo/chandler-dev

Reply via email to