forrestbot switching to plain svn command

[Tim Williams <wi...@gmail.com>]

I've got this working locally, but I use cache credentials (e.g.
keychain).  I'd personally not be comfortable putting my credentials
in clear text in a properties file.  That's the part
(deploy.svn.settings) that, I haven't worked.  So, I'll offer some
options up:

1) We can remove the credential business totally and assume that folks
work out their authentication on their own and I'll check in what I
have.

2) If someone else has an itch and really wants the credentials to be
there, I can either a) checkin what I have and they can go about
adding it in or b) we can create a JIRA ticket and I'll attach it to
that.

Any thoughts/recommendations?

Thanks,
--tim

Re: forrestbot switching to plain svn command

[David Crossley <cr...@apache.org>]

Sina K. Heshmati wrote:
> David Crossley wrote:
> > I am going ahead to just do this FOR-1198 to use svn executable
> > and auth only. I will fix docs and instructions too.
> > 
> > If someone wants to add "settings" file ability later,
> > then i will help with that. Actually i am happy that
> > it is not necessary anymore, and have given up wondering
> > what changed behind-the-scenes.
> 
> Sounds good. Thanks for applying Tim's patch [1].
> 
> [1] http://svn.apache.org/viewvc?view=revision&revision=951817

That was only part of it. There are other changes too.
See the "Subversion Commits" view at FOR-1198.

However, i am going to need to review it all
because it needs to handle the case where the
username that it is running as, is not the same
as the "svn user".

-David

Re: forrestbot switching to plain svn command

["Sina K. Heshmati" <si...@khakbaz.com>]

David Crossley wrote:
> I am going ahead to just do this FOR-1198 to use svn executable
> and auth only. I will fix docs and instructions too.
> 
> If someone wants to add "settings" file ability later,
> then i will help with that. Actually i am happy that
> it is not necessary anymore, and have given up wondering
> what changed behind-the-scenes.

Sounds good. Thanks for applying Tim's patch [1].

Sina

[1] http://svn.apache.org/viewvc?view=revision&revision=951817

Re: forrestbot switching to plain svn command

[David Crossley <cr...@apache.org>]

I am going ahead to just do this FOR-1198 to use svn executable
and auth only. I will fix docs and instructions too.

If someone wants to add "settings" file ability later,
then i will help with that. Actually i am happy that
it is not necessary anymore, and have given up wondering
what changed behind-the-scenes.

-David

Re: forrestbot switching to plain svn command

[David Crossley <cr...@apache.org>]

Sina K. Heshmati wrote:
> Tim Williams wrote:
> > Sina K. Heshmati wrote:
> >> Tim Williams wrote:
> >>
> >>> I've got this working locally, but I use cache credentials (e.g.
> >>> keychain).  I'd personally not be comfortable putting my credentials
> >>> in clear text in a properties file.  That's the part
> >>> (deploy.svn.settings) that, I haven't worked.  So, I'll offer some
> >>> options up:
> >>>
> >>> 1) We can remove the credential business totally and assume that folks
> >>> work out their authentication on their own and I'll check in what I
> >>> have.
> >>>
> >>> 2) If someone else has an itch and really wants the credentials to be
> >>> there, I can either a) checkin what I have and they can go about
> >>> adding it in or b) we can create a JIRA ticket and I'll attach it to
> >>> that.
> >>>
> >>> Any thoughts/recommendations?
> >>
> >> Maybe something like:
> >>
> >> 1) If the deploy.svn.settings file exists, use it, if not
> >> 2) Prompt user for credentials. If user input is empty,
> >> 3) Rely on ~/.subversion i.e. auth-cache.
> > 
> > Thanks Sina, I wasn't clear.  It was really whether or not people see
> > it as significant enough to be missed.  I'm happy to give pointers on
> > how to get it done but I haven't the time or inclination to do the
> > various credential-cache-clearing maneuvers required to properly test
> > it.  The code is trivial, testing this particular thing fully is the
> > burden - I'd be happy to implement it if someone else were inclined to
> > test it:)  For now, I'll create a JIRA issue, check in what I have and
> > if someone wants to build on it, they can.
> 
> Well, it should be significant to those who can't publish the site.

Yeah, i will be testing it! I am crippled at the moment.

I hope that others do test it too.

It is not just "forrestbot". Also cannot deploy any plugin
or any plugin's website.

> An issue has been created on JIRA by Tim based on this discussion:
> https://issues.apache.org/jira/browse/FOR-1198
> 
> This discussion logically follows the following thread:
> http://forrest.markmail.org/thread/ctda7nqukdv3dpy5

Thanks for linking stuff together. That is so important
for the archives.

I reckon that we can rely on the ~/.subversion/auth

These actions (forrestbot deploy and plugin deploy) are being
done by a developer who has Subversion installed.

However there is a situation where forrestbot could be
installed from a Forrest release package on a server
where Subversion might not be present.

I suggest that we note that in the documentation.

There could not be an "upgrading" problem, because
the current implementation already uses a mixture
of svn executable and Jsvn.

-David

Re: forrestbot switching to plain svn command

["Sina K. Heshmati" <si...@khakbaz.com>]

Tim Williams wrote:
> On Sat, Jun 5, 2010 at 6:59 PM, Sina K. Heshmati <si...@khakbaz.com> wrote:
>> Tim Williams wrote:
>>
>>> I've got this working locally, but I use cache credentials (e.g.
>>> keychain).  I'd personally not be comfortable putting my credentials
>>> in clear text in a properties file.  That's the part
>>> (deploy.svn.settings) that, I haven't worked.  So, I'll offer some
>>> options up:
>>>
>>> 1) We can remove the credential business totally and assume that folks
>>> work out their authentication on their own and I'll check in what I
>>> have.
>>>
>>> 2) If someone else has an itch and really wants the credentials to be
>>> there, I can either a) checkin what I have and they can go about
>>> adding it in or b) we can create a JIRA ticket and I'll attach it to
>>> that.
>>>
>>> Any thoughts/recommendations?
>>
>> Maybe something like:
>>
>> 1) If the deploy.svn.settings file exists, use it, if not
>> 2) Prompt user for credentials. If user input is empty,
>> 3) Rely on ~/.subversion i.e. auth-cache.
> 
> Thanks Sina, I wasn't clear.  It was really whether or not people see
> it as significant enough to be missed.  I'm happy to give pointers on
> how to get it done but I haven't the time or inclination to do the
> various credential-cache-clearing maneuvers required to properly test
> it.  The code is trivial, testing this particular thing fully is the
> burden - I'd be happy to implement it if someone else were inclined to
> test it:)  For now, I'll create a JIRA issue, check in what I have and
> if someone wants to build on it, they can.

Well, it should be significant to those who can't publish the site.

An issue has been created on JIRA by Tim based on this discussion:
https://issues.apache.org/jira/browse/FOR-1198

This discussion logically follows the following thread:
http://forrest.markmail.org/thread/ctda7nqukdv3dpy5

SinDoc

Re: forrestbot switching to plain svn command

[Tim Williams <wi...@gmail.com>]

On Sat, Jun 5, 2010 at 6:59 PM, Sina K. Heshmati <si...@khakbaz.com> wrote:
> Tim Williams wrote:
>
>> I've got this working locally, but I use cache credentials (e.g.
>> keychain).  I'd personally not be comfortable putting my credentials
>> in clear text in a properties file.  That's the part
>> (deploy.svn.settings) that, I haven't worked.  So, I'll offer some
>> options up:
>>
>> 1) We can remove the credential business totally and assume that folks
>> work out their authentication on their own and I'll check in what I
>> have.
>>
>> 2) If someone else has an itch and really wants the credentials to be
>> there, I can either a) checkin what I have and they can go about
>> adding it in or b) we can create a JIRA ticket and I'll attach it to
>> that.
>>
>> Any thoughts/recommendations?
>
> Maybe something like:
>
> 1) If the deploy.svn.settings file exists, use it, if not
> 2) Prompt user for credentials. If user input is empty,
> 3) Rely on ~/.subversion i.e. auth-cache.

Thanks Sina, I wasn't clear.  It was really whether or not people see
it as significant enough to be missed.  I'm happy to give pointers on
how to get it done but I haven't the time or inclination to do the
various credential-cache-clearing maneuvers required to properly test
it.  The code is trivial, testing this particular thing fully is the
burden - I'd be happy to implement it if someone else were inclined to
test it:)  For now, I'll create a JIRA issue, check in what I have and
if someone wants to build on it, they can.

--tim

RE: forrestbot switching to plain svn command

["Sina K. Heshmati" <si...@khakbaz.com>]

Tim Williams wrote:

> I've got this working locally, but I use cache credentials (e.g.
> keychain).  I'd personally not be comfortable putting my credentials
> in clear text in a properties file.  That's the part
> (deploy.svn.settings) that, I haven't worked.  So, I'll offer some
> options up:
> 
> 1) We can remove the credential business totally and assume that folks
> work out their authentication on their own and I'll check in what I
> have.
> 
> 2) If someone else has an itch and really wants the credentials to be
> there, I can either a) checkin what I have and they can go about
> adding it in or b) we can create a JIRA ticket and I'll attach it to
> that.
> 
> Any thoughts/recommendations?

Maybe something like:

1) If the deploy.svn.settings file exists, use it, if not
2) Prompt user for credentials. If user input is empty,
3) Rely on ~/.subversion i.e. auth-cache.

SinDoc