Activity
From 01/12/2010 to 02/10/2010
02/10/2010
-
10:01 PM Feature #144: hosts decorator can not accept a list object as it's argument
- I'm not sure how I feel about this -- I generally try to avoid 'morphing' function signatures except in extreme situations, and I don't see why the function signature for `@hosts` should get all warped just to save someone the trouble of doing `@h...
-
- I haven't done anything on this other than learning more about how bash completion works, myself, for unrelated reasons a few weeks ago :) What do you think's needed inside Fabric? Offhand, what we'd need is: * A CLI flag similar to `--list`...
-
07:31 PM Feature #144: hosts decorator can not accept a list object as it's argument
- For example, doing the following: ...
-
06:48 PM Feature #6: Bash/Zsh/etc completion
- I'm interested in seeing some more progress on this ticket. The scripts posted here are a good start, but I think some changes are needed from Fabric to support tab completion on arguments. Has any work been done on this lately? I'm willing to ...
-
-
- For example, the following code always errors out for me: ...
-
- Thanks for the catch, interesting bug. Should be easily fixable.
-
11:52 AM Bug #142: disconnect_all doesnt delete closed connection from connections cache
- hi, i use fabric version 1.0a. after a network.disconnect_all() the connections cache is not empty. if you call run (or something similar) again it is raise an exception: ...
-
02/09/2010
-
03:25 PM Feature #28: Allow put-style globbing with get
- This ticket is related to #140, and is automatically fixed it that is implemented.
-
- This impacts ticket #140 as well.
-
- There was only one instance in the tree now, fixed in my git repos: http://github.com/sophacles/fabric/tree/bug37
-
- Especially the `look_for_keys` argument to `connect` -- otherwise some users (see: OgMaciel on IRC) who have SSH keys but don't *want* to use them, will still get prompted for their passphrase. This should be a simple matter of adding new env v...
02/08/2010
-
- I did some work on this, available in [my git repo](http://github.com/sophacles/fabric/tree/feature140). In it I added a directory called test_scp (at the project root), which has a test fabfile and a directory structure for testing. There are a ...
-
- Popping this in the warts list, more and more folks are assuming that Python-called subtasks will preserve their own specific host list, and it definitely feels like a major wart. In the meantime point people to http://lists.gnu.org/archive/htm...
02/07/2010
-
- Something doesn't seem right here, the code at http://bpaste.net/show/3111/ seems to contradict the assertions in this report.. What am I missing here?
-
- Some initial work re my previous comment: http://github.com/sophacles/fabric/tree/bug138 Not sure what else would be needed to get everything happy. Edit: Ok, I did that before I noticed Jeff had made his comment above. Since the work was alr...
-
- Yea, that would do it. The more I think on it, though, the more I think this may be best solved by jumping straight to something that solves #4 -- a global "use this port" only makes sense in the "connecting to one and only one server" scenario...
-
- OK, thanks :) Tickets can always be reopened.
02/06/2010
-
- The current functionality of put and get are limited by only being able to place files in existing dirctories, and further (because of glob limitations), dont support deep recursion. This is fairly annoying for deployments, so I suggest using a mo...
-
- This is because everything in operations does this: connections[env.host_string] There is never any checking against env.host, env.port, and env.user. This propagates to `HostConnectionCache` where all opers are done based on normalize(...
-
- Yeah, after some more thought I've decided namespaces may actually solve this one better. Im going to withdraw my request, but reserve the right to ask again later :P.
-
- Put this code, plus updated docs in [my fork](http://github.com/sophacles/fabric/tree/bug137).
-
- Fixed in my branch at: http://github.com/sophacles/fabric/tree/bug110
-
- I should note, I kept the abort() message too-- displaying help + a wtf is nice.
-
- Simple fix for this in my github fork here: http://github.com/sophacles/fabric/tree/bug75
02/05/2010
-
- That's true, we could still do (yea, ugly) "static typing", but push it up to the task definition instead of leaving it on the command line as my earlier rant had mentioned. This makes more sense anyway, now that I think about it. ALso, I'll li...
-
- Just a thought on this, perhaps a arg_types decorator, it feels ugly but also feels like the right starting point. Example: @arg_types(BoolArg('x'), IntArg('y','z')) def task(x, y, z): if x: do_something(y*z) ...
-
- Good points; I'm convinced. Thanks for taking the time to elaborate.
-
- False alarm -- looks like it was dropped packets on the remote end, so a timeout error is actually correct.
-
- Also from sm0ke in IRC -- the problems in #138 were exacerbated by a timeout error message being shown instead of connection refused. I assume this is due to a too-wide Paramiko exception, but should double check this assumption. We may be maki...
-
- User sm0ke in IRC says he had a problem where his remote system had a nonstandard SSH port, and his use of `env.port` was ignored -- only specifying the port number in the host string worked. I seem to recall this being fixed before but don't s...
-
- This actually came up twice for me. Once in a task which does some mucking around with files kept in subversion. The final step of this task is to do a checkin. One of my parmeters from the commandline is commit_message, which is just a string. Th...
-
- Sorry to be so interrogative :) Is there a reason commas are necessary in the values given? I.e. is there a reason you cannot have your task split on something other than a comma, like a semicolon? That's generally what I do. This might be rela...
-
- I don't doubt that this is useful for you, but can you provide a realistic use case example? I'm having a hard time thinking of where this is useful generally, it feels like it may be solving something that can already be accomplished in another m...
02/04/2010
-
- Actually line 13 of the paste should be: ...
-
- It sometimes comes up that a comma will be needed as part of an argument passed on the command line. In such a case an escaped comma option would be nice. There is a solution at http://paste.pocoo.org/show/174044/ if you want that.
-
- I would like to be able have the ability to break my fabric tasks into sub-tasks per directory, yet executable with one overarching task call. So when i am in `project_root/` I could call `fab task1`. fabric would then do the actions specified in...
-
- I came up with a way of doing this host metadata that works pretty well for the simple cases. The relevant code bits are here http://paste.pocoo.org/show/173964/ however there are some ideas that I didn't have time to implement yet -- namely that ...
02/01/2010
-
10:30 PM Feature #76: Use decorator to define tasks
- I just found the use case for both a `@task` decorator and `__all__`. I've started work on some reusable tasks for various things. I'm relying heavily on namespaced tasks (`module.some.task`) in the [RabbitMQ][] tasks. I have a `ctl` task at ...
01/24/2010
-
- Carrying over the conversation from #135, the core `logging` module is what should be used for all output. That gets logging to file for free as a side-effect, but more importantly provides a uniform way to handle formatting output.
-
-
- Python provides an excellent [`logging`][logging] module for generating log output. The vast majority of output generated by Fabric while running can be considered log material. Fabric should employ `logging` for it's main output. It would simp...
-
- I took at stab at the decorator task definitions too. Changes are in my repo at commit [3676810][]. I took a different approach for handling the registration than Andy's version. It does the registration by looking for something that is a subcl...
01/22/2010
-
- Yea, my thoughts were alone those lines: 1.0 looks for (depending on how we do it) `__all__` and/or decorated tasks, and if it finds any, then only those are used; if it finds none, it falls back on the current behavior. And yes this would prob...
01/21/2010
-
- Wonder what your thoughts are on this with existing fabfiles for BC? My thoughts would be to introduce theses like my __all__ change. If one variable is a Task, they all have to be, but if none are specified, then treat it like a normal fabfile....
01/17/2010
-
- I like this, with one caveat. I'd make the optional string a regular expression.
01/14/2010
-
07:09 PM Feature #56: Add namespacing or dot notation
- I agree with your point, however I feel fabric CLI parameters should have all the power deciding what to include. It's how Fabric provides flexibility to different situations and environments. Code only provides the tools. @Jeff, sorry about ...
01/12/2010
-
- When using large fabfiles, one may wish to search for or filter task names. E.g. you know you have some relating to Python but aren't entirely sure and want to do something like `fab --list "py*"`. Benefit of adding this to `-l`/`--list` is tha...
Also available in: Atom