Prior to 2.1, Capistrano would always allocate a "pseudo-tty" (or
pty) for every command it ran. This allowed commands that did
windowing stuff (e.g., via curses and friends) to query things like
the size of their window and so forth. However, it also meant that
when you ran a command via Capistrano, it would run in an environment
where your .profile (or .bashrc, or whatever) was not automatically
loaded.
With cap 2.1, I removed the auto-allocation of the pty, with the
effect that now people's .profile files will be autoloaded for every
command.
Alas, the side effect is that some commands, in some environments,
will apparently not prompt or run correctly when executed in the
absence of a pty. :( I may need to do a follow-up release soon that
makes the pty allocation the default, and allows you to disable it if
you want.
- Jamis
On Oct 16, 2007, at 7:19 PM, jared wrote:
> Added that line and everything ran fine, thanks a lot for the
> response.
> Is this something I should add to all my recipes, or can you offer
> some wisdom on why this happened/how to avoid it in the future?
> Thanks Again!
> On Oct 16, 2:30 pm, Jamis Buck <ja...@37signals.com> wrote:
>> See if adding the following line makes any difference:
>> default_run_options[:pty] = true
>> - Jamis
>> On Oct 16, 2007, at 3:28 PM, jared wrote:
>>> I just upgraded to cap 2, and I'm getting the following error when
>>> trying to deploy:
>>> $ cap staging deploy
>>> * executing `staging'
>>> triggering start callbacks for `deploy'
>>> * executing `deploy'
>>> * executing `deploy:update'
>>> ** transaction: start
>>> * executing `deploy:update_code'
>>> * executing "svn checkout -q -r392 svn+ssh://myproject.com/repos/
>>> myproject /var/www/apps/dev.myproject.com/releases/20071016185844 &&
>>> (echo 392 > /var/www/apps/dev.myproject.com/releases/20071016185844/
>>> REVISION)"
>>> servers: ["myproject.com"]
>>> [myproject.com] executing command
>>> ** [err] Host key verification failed.
>>> ** [err] svn: Connection closed unexpectedly
>>> command finished
>>> *** [deploy:update_code] rolling back
>>> * executing "rm -rf /var/www/apps/dev.myproject.com/releases/
>>> 20071016185844; true"
>>> servers: ["myproject.com"]
>>> [myproject.com] executing command
>>> command finished
>>> command "svn checkout -q -r392 svn+ssh://myproject.com/repos/
>>> myproject /var/www/apps/dev.myproject.com/releases/20071016185844 &&
>>> (echo 392 > /var/www/apps/dev.myproject.com/releases/20071016185844/
>>> REVISION)" failed on myproject.com
>>> So it seems there's a problem with my ssh login, but
>>> 1) I'm not telling it to use an ssh key, it seems like it should be
>>> prompting me for a password, so I tried adding ssh_options[:keys] =
>>> %w(/Users/[myusername]/.ssh/id_rsa), but got the same error. Next I
>>> tried deleting my authorized_keys entry on the remote server, after
>>> which I did get prompted for a password twice, but the same error
>>> comes up.
>>> 2) I have no problem logging in to ssh and running the checkout
>>> command from the console
>>> What am I missing?
>>> Thanks.
>> smime.p7s
>> 3KDownload
> --~--~---------~--~----~------------~-------~--~----~
> To unsubscribe from this group, send email to capistrano-
> unsubscribe@googlegroups.com
> For more options, visit this group at http://groups.google.com/
> group/capistrano
> -~----------~----~----~----~------~----~------~--~---
