Mozilla's policy on authentication with versioning systems
Alexandru Szasz
I know that I've read once that other authentication forms than ssh
keys are a no go, but I don't remember reading why.
Could someone tell me why is the authentication with ssh key the only
one available ?
I'm talking about the Mozilla's versioning systems, CVS, SVN, upcoming hg.
It's hard to use a SSH key on a windows system (me = Linux user), but
the real problem comes when a localization tool will want to do some
automation.
For example the tool I'm developing, Narro, has the supreme goal of
publishing translations through the push of a button, and that would
be almost impossible if the versioning system uses SSH keys; unless I
ignore the privacy of a private key and allow uploading such a thing.
Has anyone at Mozilla considerd Transifex ( http://transifex.org/ ) ?
It's in python, it's actively developed and would save Mozilla the
trouble of giving HG,CVS,SVN accounts to localizers that don't really
need such a thing, nor working with the tools. For those who don't
know, Transifex is a web interface between localizers and various
versioning systems. You can upload a new file version via web instead
of committing through command line or some software.
--
Alexandru Szasz
Axel Hecht
I talked to Dimitris about transifex at foss.in. That aside.
You're not asking the right question. The question is not about
authentication, the question is about whether we'd open up our source
code repositories to remote tools to check in stuff.
I don't think that we did that so far, and I don't know what it'd take
to do that.
Axel
Alexandru Szasz
> Alexandru Szasz wrote:
>> Hi,
>>
>> I know that I've read once that other authentication forms than ssh
>> keys are a no go, but I don't remember reading why.
>> Could someone tell me why is the authentication with ssh key the only
>> one available ?
>> I'm talking about the Mozilla's versioning systems, CVS, SVN, upcoming hg.
>>
>> It's hard to use a SSH key on a windows system (me = Linux user), but
>> the real problem comes when a localization tool will want to do some
>> automation.
>> For example the tool I'm developing, Narro, has the supreme goal of
>> publishing translations through the push of a button, and that would
>> be almost impossible if the versioning system uses SSH keys; unless I
>> ignore the privacy of a private key and allow uploading such a thing.
>>
>> Has anyone at Mozilla considerd Transifex ( http://transifex.org/ ) ?
>> It's in python, it's actively developed and would save Mozilla the
>> trouble of giving HG,CVS,SVN accounts to localizers that don't really
>> need such a thing, nor working with the tools. For those who don't
>> know, Transifex is a web interface between localizers and various
>> versioning systems. You can upload a new file version via web instead
>> of committing through command line or some software.
>>
>
> I talked to Dimitris about transifex at foss.in. That aside.
Is there any chance to adopt it ?
>
> You're not asking the right question. The question is not about
> authentication, the question is about whether we'd open up our source
> code repositories to remote tools to check in stuff.
>
The question is about authentication. I could ask localizers to upload
their private ssh key to commit something right now and probably most
of them would't care about the privacy of a ssh key since many of them
use this key only for Mozilla.
What I would like to know is what makes key based authentication
preferable to password based authentication in Mozillaland ?
--
Alexandru Szasz
Jesper Kristensen
> What I would like to know is what makes key based authentication
> preferable to password based authentication in Mozillaland ?
Mozilla does not use ssh only. You can commit to SVN using https and
your LDAP password. I don't know about hg.