Re: [Fab-user] status of "occasional droped lines of output" bug 32

[Jeff Forcier]

Before I go into my reply to Steve below, an announcement: I just
pushed what is hopefully a fix to this issue, to both the 0.9 and
master branches. See http://code.fabfile.org/issues/show/32 for
details.

I'd greatly appreciate it if Xinan, Carl and anyone else with this
issue could confirm that it fixes the problem for them (and for anyone
else to live on the edge and grab it to confirm it doesn't cause other
problems...)

On to my reply...


On Sat, Aug 8, 2009 at 12:14 AM, Steve Steiner
(listsin)<address@hidden> wrote:

> Cool idea, though.  Finding "the fabric installation dir where I'm to run my
> dir/ls command" is probably not going to be that much fun ;-0.

How so? I was just thinking of using e.g. os.path.dirname(__file__).
Since we're SSHing to localhost and running the test from localhost,
that ought to get us what we need, unless I'm missing something.

> IOW, a test against a known host on the Internet lets us test the whole
> stack, against a known quantity, and gives the fewest variables in all
> tests.

Valid points. After thinking about it, I'm not sure I'm happy with the
idea of opening up even a limited user to the public, however =/ call
me paranoid but it's just asking for me to get rooted because of some
silly vulnerability in "ls" or whatever the user is given access to.

I know that in the past there used to be free "get an SSH shell for
doing IRC or whatever" hosts out there; if those still exist they
might potentially come in handy for this. Not sure I have the $$$ for
another full-fledged "sandbox" VPS for this purpose, which would of
course be another solution. (I'd rather not ask people for handouts on
this either, but that's not to say I would turn down a
guaranteed-to-be-valid-for-the-long-term offer :))

> Writing mock objects for easily created real-life objects has never, in my
> experience, been worth the effort.  This is especially true now when
> real-life resources are so cheap and easy to set up.

I agree in principle; for *now* I was (surprisingly) able to easily
test this particular issue by mocking the SSH channel object, but I'm
definitely still open to the "live test" setup once it becomes
necessary. Once the "easy to mock" turns into "kind of a pain to mock"
or "oh god I hate mocking" :)

Regards,
Jeff