Activity
From 05/24/2010 to 06/22/2010
06/22/2010
-
05:07 PM Revision 52a66584fde759b49cefeee541d045abcba79973 (7-prompt-detection, master, ...): Slight reorg...
- Slight reorg of _execute to be more self-contained, easier to test
-
- Spinoff from #183, though that's mostly not important. Need to test real, honest to God SSH level stuff, i.e. `run()` and its constituent underpinnings, if we're to have any chance of a serious test suite. There are, as I see it, 3 ways to d...
-
02:44 PM Feature #184: allow local_dir to be a list in rsync
- It's only a little extra to the local_dir line. [github/goosemo/fabric](http://github.com/goosemo/fabric/commit/0f9a1c5e46692004e8c39f7f56dc817bb9fb881b)
-
- I think it should perhaps mention that the rsync_project() is a pretty bare wrapper around the rsync command. Feel free to poke at the code to see some of this: http://github.com/bitprophet/fabric/blob/master/fabric/contrib/project.py ; since it i...
-
11:40 AM Feature #184: allow local_dir to be a list in rsync
- I needed rsyncing only a subset of the current project so subdirs and files A B,C, D etc. The wording in the documentation implies that there is only one local_dir, and does not mention this workaround (which is more or less the same as writing t...
-
11:31 AM Feature #140: Recursive put() and get()
- Just checked in some "get rid of random debugging print statements" fixes, and verified that pulls from multiple hosts do directories in a sane way -- specifically creating a directory of $HOSTNAME and pulling to that location.
-
- What were you looking to be able to do? Right now it is like that because it is a straight use of rsync form the cli. You can simply use: ...
-
- This was actually tackled as part of #7. I liked gevent pretty well, but a major kink in the works (ironically) is that while coroutine based stuff seemed to at least partially work for the remote output streams, I could not find any way to handle...
-
- rsync allows copying multiple local dirs to the target, please support this by allowing local_dir to be an iterable as excludes Now this effect can be achieved by making local_dir='dir1 dir2 dir3' but feels unclean :)
06/21/2010
-
- Fixed that issue, but found 1-2 others, for which I've made tickets. Also realized that the "no prefixes" issue is really part of another recently opened ticket, and just commented on that one instead. Going to close this ticket now to keep it ...
-
- In the #7 feature branch, I noticed that when testing the prompt features, it seems that all output post-prompt-handling is no longer captured. This should be easy enough to fix, but the real reason for this ticket is to come up with a decent w...
-
- Print sudo prompts, try-again notices if necessary. Re #7
-
- Keep in mind that a fully empty prefix should also be easy to effect, though offhand that should simply be an empty-string prefix setting, so not a big deal. For example, now that #7 is basically done, fully interactive stuff is going to want p...
-
- When rewriting the I/O loop I had it print the line prefix after any line ending, meaning `\n` OR `\r`. I believe I did this because even during testing I ran into cases where a `\r` was present without a `\n`, though I'm not sure. It may have sim...
-
- I'm dumb; should just be possible to detect whether the stream the prompt is detected in, is being hidden, and print it manually if so.
-
- Yup; needs more work. Right now the main issue is, as mentioned, prompts being hidden by low level subroutines using `hide`. When we printed the prompt ourselves this wasn't a problem, but now it is. Gonna have to think about this; it's the age-ol...
-
- Fix couple of stupid mistakes in pty stuff
-
- Fix missing comma
-
- Additional expansion of interaction docs
-
- Fill out interactivity docs re #7
-
- Change --pty to be --no-pty re #7
-
- Reorder, expand backwards compat note re: #7
-
- Update env var docs re: #7
-
- Pty env var and kwarg now disable when given. IOW, now that the default has swapped from False to True, we need to change our test such that if either is False, we don't use a pty.
-
- More reshuffling of 1.0 changelog
-
- Merge master into 7-prompt-detection, add changelog entry
-
- Re #7, add/tweak API docs
-
- Re #7, add Windows stdin bytewise support.
-
- Test tweaks for non-git environments. Why you'd be developing Fabric without Git installed, I have no idea. I guess for those users who like grabbing tarballs and running the test suite (valid, just unusual).
-
- Update tests to pass after recent work
-
- Re #7, some basic code reorg
-
- Re #7, tweak function signatures
-
- Add first draft of open_shell(). (Was going to be named interact_with() but this is more general.) Re #7
-
- Comment
-
- Update again_prompt to correctly handle pty or lack thereof
-
- Don't display prompt when IO loop is prompting for password
-
- Reflow password capturing to work with new loop style
-
- Minor refactor/tweak
-
- Print initial prefix at beginning of output.
-
- Fix out/err prefixes to incldue host string again
-
- Got stdin working again, for the most part.
-
- Handful of minor tweaks to user-facing output
-
- Back to threads for I/O looping. Semi-works.
-
- Make trailing newline honor output controls
-
- Redo _write to honor output controls again.
-
- Copy/modify password memory feature from output_thread
-
- Add trailing newline after I/O loop
-
- Add env vars for echoing and stderr combining
-
- Abstract bytewise buffering to be context manager. Re #7
-
- Reorganize IO loop a bit, add prefix
-
- First experimental steps with select() etc
-
- Add env.local_user for easy local username access.
-
- Sort initial env dict
-
09:34 AM Bug #179: run behaviour changed in 0.9.1
- Nope, I used the simple fab in the initial report. Just to narrow it down :) Thanks for the rapid responses
-
- Are you using the new-in-master `with path():` context manager, in your master version of the code? If not, that's likely where the bug is; I don't think I was intending it to add anything to the command being run unless the user had explicitly...
-
- The output on master: [krusty] Executing task 'migrate' [krusty] run: /bin/bash -l -c "cd /home/citylive//django_projects/mobilife && PATH=\$PATH:\"/home/citylive/\" . /home/citylive//env/mvne/bin/activate; python manage.py syncdb --noin...
06/20/2010
-
- We're done! Now to merge into master, preferably with some sort of advance warning for the bleeding-edge users. I'll probably just use the ML but I'd prefer to get a blog set up instead, probably a clone or refactor of the one I'm building for my ...
-
- Additional expansion of interaction docs
-
- Fill out interactivity docs re #7
-
- Change --pty to be --no-pty re #7
-
- Reorder, expand backwards compat note re: #7
-
- Update env var docs re: #7
-
- Re the duplicate ticket #181, while an env var should allow use of `with settings():`, consider making a convenience `with sudo():` as well (though we'd need to use some other name, in which case it may not be worth it -- think about it).
-
-
- Pty env var and kwarg now disable when given. IOW, now that the default has swapped from False to True, we need to change our test such that if either is False, we don't use a pty.
-
12:19 PM Feature #181: Sudo context and/or decorator
- I wonder if it is technically possible but I'd like to set the "run with sudo" globally by using context or decorator. Or more precise, I'd like to drop the sudo parameter from *all* the functions, just because it is needed there for consistency. ...
06/19/2010
-
- More reshuffling of 1.0 changelog
-
- Just-woke-up rambling: Ticket #7 is basically done, but unfortunately it's not the more elegant of the two -- the tutorial would need to show users how to push a private key to the remote end in such a way that the remote SSH picks it up (at whic...
06/18/2010
-
- Merge master into 7-prompt-detection, add changelog entry
-
- Reformat 1.0 changelog, add issue links
-
- Re #7, add/tweak API docs
-
- Reformat 1.0 changelog, add issue links
-
- Considering this done for now.
-
- This was fixed in master/1.0 a short while ago -- single quotes are now escaped.
-
- Re #7, add/tweak API docs
-
- Re #7, tweak function signatures
-
- Not gonna bother with the sudo version, this has such limited utility anyways, and furthermore, now that #7 is in, one can simply call sudo directly. The `sudo` operation is really only there to make the password memory easier at this point. So...
-
- Test tweaks for non-git environments. Why you'd be developing Fabric without Git installed, I have no idea. I guess for those users who like grabbing tarballs and running the test suite (valid, just unusual).
-
- Update tests to pass after recent work
-
- Re #7, some basic code reorg
-
- Just a note, the work done over in #7 now means that remote stdout/stderr are printed on a byte-by-byte basis instead of line-by-line (fabfile-local print statements/interactions with `sys.stdout` remain line-buffered as usual). This may impact an...
-
- While it requires 3 or 4 more `win32` testing `if` statements (gross) I am happy to report that at least under ActiveState's Python 2.6.x, the `msvcrt` module works as advertised. Furthermore, Cygwin seems to be happy with the Unixisms as usual.
-
- Re #7, add Windows stdin bytewise support.
-
- Pressure from #158, and the fact that I think it should be easy enough for us to host a tgz of Paramiko 1.7.4 for edge case users, means I am thinking seriously of nuking our vendorized Paramiko for the 0.9.2 release. New users are relatively h...
-
- Merge branch 'master' into 7-prompt-detection
-
- Applied in changeset commit:dae39c20f3259c006b680b1d02d7c70a82f5d124.
-
- Consume stderr to hide lack-of-git error messages. Also hide empty parens in same situation. Fixes #180
-
- (Note for anyone too lazy to click over to #123 -- this has just been addressed by removing the cygwin test. Landed in all branches.)
-
-
- Merge branch 'master' into 7-prompt-detection
-
- Yup, ActiveState reports `win32`, and I can confirm that Cygwin falls down quite hard with the way things currently stand. Also confirming that simply nuking the cygwin test makes things work just fine (certainly, the two places we use the flag bo...
-
- Merge branch '0.9'
-
- Remove Cygwin from win32 flag test. Re #123
-
- If the user does not have Git installed / is not working off a git cloned version of the source, `fab --version` works but spits out "hey, what's this 'git' thing?" errors. Fix this.
-
- Adding note re #146. In testing out some stuff for #7 I am also trying to tackle this issue (though I'll be landing these changes in 0.9.x and only then porting up to master and the #7 feature branch). I have Cygwin and its Python installed ...
-
- This looks to be the same issue as #123, actually.
-
- Oh, another thing to try is to add `--show=debug` to your `fab` invocation, or wrap the `run` in question in a `with show('debug'):` block -- then see what it reports as the actual command being run. (Nuke the "env | grep VIRTUAL_ENV" stuff too --...
-
- That is pretty odd, then. Thanks for testing in 0.9.1 -- that will narrow down the list of changes I need to scan to see what might have caused such an issue. I'll see what I can find. In the meantime, would you be able to share the rest of you...
06/17/2010
-
- * Added first draft of `open_shell` (née `interact_with` -- realized there was no point *assuming* that it would be used with an initial command, and as such the name made less sense as -is). Works pretty well for me. * Occasional hiccups with...
-
- Add first draft of open_shell(). (Was going to be named interact_with() but this is more general.) Re #7
-
- Strange thing is that it works on 0.9.0 & 0.9.1 $ fab krusty migrate [krusty] Executing task 'migrate' [krusty] run: . /home/citylive//env/mvne/bin/activate; env | grep VIRTUAL_ENV; python manage.py syncdb --noinput [krusty] ou...
-
- Comment
-
- Unable to recreate the issues with prompts showing up mid-line. I recall it being because of stdout and stderr, e.g. a prompt is printed to stderr before any stdout has appeared, then the user's input is echoed on stdout, triggering the stdout pro...
-
- Yea, master has a number of changes, though offhand I don't see why any would cause your error there. (Given that it's actually an error from your manage.py, there may be something more subtle going on.) Any chance you could paste more of the o...
-
- Right, sorry. 1. It seems that the `virtualenv` is not activated. So in my case it can't find the [krusty] err: Traceback (most recent call last): [krusty] err: File "manage.py", line 2, in <module> [krusty] err: ...
-
- Did you accidentally submit the form before finishing? It sure looks that way :) You never said what broke! Based on what you entered you may have been relying on a bug in `cd` where it would not correctly replace the path if given an absolute ...
-
- Great, I am able to totally hide the `getpass` prompt, *and* I was also able to address the two issues mentioned in the previous comment, so the prompting now behaves pretty well in my testing. There are edge case issues with `prompt_for_passwo...
-
- Don't display prompt when IO loop is prompting for password
-
- Reflow password capturing to work with new loop style
-
- Update again_prompt to correctly handle pty or lack thereof
-
- Minor refactor/tweak
-
- When upgrading from 0.9.0 to 0.9.1 some of my scripts stopped working. The idea is to activate a virtualenv and execute some command (e.g. syncdb for Django) def syncdb() with cd(env.path): run('. /home/gvangool/en...
06/16/2010
-
- In testing this I realized there was a pre-existing bug, namely that with pty=True, the "Sorry try again" text ends with \r\n instead of just \n. WIth pty=False it's just \n. Just a note -- should be easy enough to test for either situation and th...
-
- Re: passwords: also need to figure out if there is a nice way to avoid double-prompting -- i.e. right now, because `prompt_for_password` prints its own prompt, the user will see the real prompt and then the Python level prompt. That's a bit too fa...
06/15/2010
-
- Turns out password memory is actually not all that difficult to copy over. Had to make some modifications due to the different data structures involved and the different algorithm, but it still seems to work in the base case and is (as with the re...
-
- Copy/modify password memory feature from output_thread
-
- Able to run my reasonably thorough setup tasks without falling over; in fact, it even does curses reasonably well! (One drawback of using a pty -- some apps may now be harder to script because they'll assume they're being used by a person, such as...
-
- Fix out/err prefixes to incldue host string again
-
- A little proof of it working: [sudo adduser -- without sudo(), and with stdin!](https://gist.github.com/a7f5cbfc9a65ecc32c53)
-
- Another idea re: prefixes -- since the default behavior is intended to be pty on, thus users will ALWAYS get the "out" prefix when typing into any non-password prompts -- is to possibly have communication between the stdin and stdout threads such ...
-
- Have thread solution basically working; certainly it appears to avoid both the crashing and the edge case of missed output, which is great. One remaining issue in my testing is that typing (i.e. sending stdin) results in an output prefix being ...
-
- Got stdin working again, for the most part.
-
- Handful of minor tweaks to user-facing output
-
- Back to threads for I/O looping. Semi-works.
06/14/2010
-
- Played with gevent some; I like its interface but I still run into issues with `select` somehow blocking everything anyways (since `select` is the only way I know of to get nonblocking stdin; was running it in a simple test program alongside anoth...
-
- Side note: in writing a simple practical test for the edge case that's prompting me to move away from the current `select` approach (it calls `run` over and over until it gets an empty result back) I'm getting intermittent errors in `paramiko.tran...
-
- Merge branch 'master' into 7-prompt-detection
-
- Add fastprint back to API module
-
- Hm, Kamaelia's site took a major downturn since I last saw it, it's got almost no info, lots of literal placeholder text, API docs are all HTTP 500s...gross. Saw mention that they're moving servers, but even so, that's unfortunate. The info I did ...
-
- Today's research notes: * Raw coroutines don't seem to map to what we need to do very well; that and `cogen` seems relatively new/poorly documented. * So the only way to handle this that I can wrap my head around is still a "thread like" appro...
-
- According to [the Popen docs](http://docs.python.org/library/subprocess.html#subprocess.Popen) (see 1st para after 2nd note block) one can force `subprocess` to use a specific shell instead of the system default (which may vary) by use of the `exe...
-
- Make trailing newline honor output controls
-
- Redo _write to honor output controls again.
06/13/2010
-
- Gave option 2 the old college try, no dice. It *seems* like it should be possible -- `Channel.fileno` actually explicitly returns a "combo" pipe, so it ought to be feasible to return the two pipes distinctly (the differentiator is which one is pas...
-
- Encountering edge case issues with the loop termination again; basically, it's the old issue where the "poll" methods Paramiko provides return True while there is still data to be read in the pipeline, and this occurs intermittently. So `channel.e...
-
- Print initial prefix at beginning of output.
-
- Add trailing newline after I/O loop
-
- Abstract bytewise buffering to be context manager. Re #7
-
- Add env vars for echoing and stderr combining
-
- Reorganize IO loop a bit, add prefix
-
- First experimental steps with select() etc
-
- Add env vars for echoing and stderr combining
-
- Merge branch '7-prompt-detection' of fabfile.org:fabric into 7-prompt-detection Conflicts: fabric/operations.py
-
- Abstract bytewise buffering to be context manager. Re #7
06/12/2010
-
- Insane, amazing, thought-provoking, tear-jerking total rewrite of ticket description. Goes into detail on how vanilla SSH behaves, compiles info from comments up till this one, and comes up with an actionable TODO list for the feature branch.
-
- Spinoff of a note from #7 which, for now, is too "big" to be included in that effort. From a quick look at the [pexpect](http://www.noah.org/wiki/Pexpect#API_Documentation) library, it's not suitable for use with Fabric since it handles the IO ...
06/09/2010
-
- Minor issue with keeping/allowing access to the bytewise buffering, is that a final newline results in a final, empty prefix being printed. Not horrible, and technically accurate (insofar as there was another "line" from the remote end, just an em...
06/08/2010
-
- Reorganize IO loop a bit, add prefix
06/07/2010
-
05:58 PM Feature #176: Add a pty option in contrib.files.upload_template function args
- As contrib.files.upload_template finally relay on run or sudo command, contrib.files.upload_template should have pty arg too...
06/06/2010
-
- Just updated, let me know how it looks to you.
-
- Re #151, minor tweak to changelog
-
- Re #151, add fastprint back in.
-
- Nuke note re: Sphinx and \n -- 0.6.6 fixed it.
-
- I agree, it was a mistake to nuke fastprint simply because it's *possible* to get the same result with `puts`. Wonder if we can come up with a slightly better name than "fastprint" however, though nothing springs to mind.
06/05/2010
-
02:33 PM Feature #151: Make fabfile print() statements controllable via output controls
- Better yet, allow it to work as `puts` with but with different defaults: ...
-
- The current implementation of this has an issue. For anyone who's relied on fastprint() as a quick way to flush a line out to `sys.stdout` without a line-ending, you now have to change the code from `fastprint(text)` to `puts(text, end='', flush=...
06/04/2010
-
- Let's compromise and say 0.9.3, 0.9.2 is already pretty heavily loaded. (I will be trying to do releases much more often, though -- it definitely won't be another 6 months from now!)
-
- First experimental steps with select() etc
-
- Id like to suggest putting this in 0.9.2 -- it is completely backwards compatible, and doesn't introduce new concepts.
-
- May also be worthwhile to have `run` and `sudo` do the `exec_command` interactivity with caveats -- i.e. stdout/stderr will be combined and stdin will be echoed always, but if that works for your use case, then the functionality is there. (Or, bas...
-
12:20 PM Feature #175: contrib.select() function
- See https://gist.github.com/2f46d4521338d5581f4b for examples. Add a select() function that prints a numbered dictionary and prompts for the user's selection. Returns the dictionary key upon selection.
-
- Took another stab at the `invoke_shell` stuff, got a bit farther, but it's still very wacky (insofar as I seem to be seeing *double* the echo I expect from my inputs, and thus have to "eat" even more than usual) and more importantly I realize ther...
-
- Glad to hear it!
-
11:44 AM Bug #174: I can not get fabric installed
- > Unfortunately, Fabric is not yet supported on Python 3 in any way, so you’ll have to use Python 2.5 or 2.6. Yes, thank you. You have correctly identified the actual problem. on Gentoo, to fix it, I did: 1) select python 2.6 via eselect 2...
-
- It looks like that server in question is using Python 3.1 (or at least, your easy_install is defaulting to 3.1, even if 2.5/2.6 is installed.) Unfortunately, Fabric is not yet supported on Python 3 in any way, so you'll have to use Python 2.5 o...
-
- Hi, On all servers it installed well, but now, on one of them I get error (will attach as text file).
06/03/2010
-
- Ticket #80 doesn't seem worth the effort at this time. Focus on finding a way to handle password-safe stdin echoing with `exec_command` instead, or if it's not possible, figure out whether the security of not echoing passwords/etc is worth spendin...
-
- Yea, this isn't going to work. Even just sentinel detection is extremely messy and error prone and complicated; given that we have to use a pty and thus can't see distinct stdout/stderr, the only remaining benefits are A) shell state and B) passwo...
-
- Committing some sorta-working but mostly-still-broken attempts at sentinel detection
-
- Bumping this to 0.9.2 for now, has potential to be a good new backwards-compatible feature to add.
-
- Doesn't seem like the non-pty approach is feasible, it's still a shell, meaning it won't exit until told to exit, and now we don't even have the shell prompt as a sentinel for determining when stuff has finished (and no, the other functions for ex...
-
- One issue with this is how to deal with logic: what if one's fabfile takes different paths depending on the results of a `run` call? (Which is kind of the point of modern Fabric's execution model, in fact.) We could add a "dry_run_return" sort ...
-
-
- Re #173, add to changelog
-
- Cherry-picked as commit:99aecd72 -- thanks!
-
- Merge branch '0.9'
- 09:55 AM Revision 99aecd72ccb1ddf530a4d1b71b6f8a6e3c4d1cc7 (master, 0.9, ...): fabric is a tool, not a fool!
- fabric is a tool, not a fool!
-
- Wow, I wonder how long that's been there? :( Looks like quite a while. Thanks for the catch!
06/02/2010
-
11:41 PM Bug #173: readme typo
- from the README: "In addition to use via the ``fab`` **fool**, Fabric's components may be imported into other Python code, providing a Pythonic interface to the SSH protocol suite at a higher level than that provided by e.g. Paramiko (which ...
-
- Have invoke_shell hooked in, need to experiment more
06/01/2010
-
- OK, so here's a semi pitfall, the shell doesn't behave *as* well unless a pty is set up (using the `SSHClient` `invoke_shell` automatically sets up a pty; using the `Channel` one does not). Unfortunately, since ptys are pseudo-*terminals* and thus...
-
- Poking at this now. Also updated description with a semi TODO list. Finally, noting that one item that hasn't been considered previously is non command related output such as login banners, console messages (if the remote system is ever set up ...
-
- Poking around at the `invoke_shell` approach now. Will be documenting it in #80 for now and report back here when I have a decision as to whether it's feasible for general use.
-
- Experimenting with invoke_shell. * No more shell or pty args as they no longer apply * Simple run (remainder) task works again after those changes. * No dropping out of shell, or exec status, etc, yet. * No testing of sudo yet.
-
- Have line-by-line input semi working, but worthless without ability to see non-newlined input from remote end
-
- Minor tweaks
-
- Partial solution in place, needs more work
-
- Re #75, tweak to work with master's `--` feature
-
- Partial solution in place, needs more work
-
- Have line-by-line input semi working, but worthless without ability to see non-newlined input from remote end
-
- Fix post-merge bug in contains
05/29/2010
-
- Couple more notes from before: * I presume the issues with crossed streams is largely because at least one stream, in normal terminal usage, is line-buffered (I believe stdout) so char-buffering isn't ever done by actual terminals. Consider lin...
05/28/2010
-
- Merge branch '0.9' Conflicts: docs/_templates/page.html fabfile.py fabric/version.py
-
-
- Tweak push_docs to make docs dir if it doesn't exist.
-
- So, naturally I neglected to consider the inverse, old docs pointing people to the newest docs. I think that'll definitely require a more generic solution than modifying Sphinx templates -- I'd have to check out the 0.9.0 version of the docs (or a...
-
- Cut 0.9.1 final
-
- As long as I update the Apache config, the 1.0 docs don't need changing, they just point to `/0.9/` which will always point to the latest release. Bumping this to 0.9.2 now; hopefully it'll get tackled sooner than that if I can get back into th...
-
- Reorganize, add more links to changelog
-
- Re #90, add static link to 0.9.1 in Sphinx header
-
- Should get an avatar for it; either reuse the old Fabric SVG logo (probably as a JPG/PNG) or come up with a new logo somehow. Maybe something silly like just monospaced text saying "$ fab" or "fabfile" or just "Fabric"?
-
- Fix erroneous merge which broke package fabfiles
05/27/2010
-
- Merge branch '0.9'
-
- Remove FAQ from manifest
-
- I install Fabric via Pip pretty frequently and it works fine for me. Your invocation looks accurate, however -- did you ever get it working? I'm hoping it was just a temporary Github glitch. Certainly, that exact invocation works for me right now.
-
- I'm not sure if this is something I want to live in actual contrib or not. Linking it to the "share unofficial snippets" ticket for now. I'll need to give your code a more thorough review later once that option is available, so I can decide where ...
-
- Bumping this to 0.9.3, actually -- I think what'd be good is to have 0.9.2 be another round of 0.9 specific bugfixes, then 0.9.3 be the backporting, and hopefully the last significant 0.9.x release.
-
- Add Python 2.5/2.6 trove classifiers re #145
-
- This is now blocking release of 0.9.1 -- I don't want to hide the 0.9.0 docs when I push up 0.9.1, and I'll have to change the 0.9 pointer to 0.9.1, so...yea. Don't really care whether this is handled manually in the Sphinx header include or if...
-
- Dropping priority on this since an FAQ has been added. (Also adding explicit link to #86.)
-
- Merge branch '0.9' Conflicts: AUTHORS docs/usage/fab.rst fabric/context_managers.py fabric/contrib/files.py fabric/operations.py
-
- Added versionadded/versionchanged notes
-
- A 1.0-specific change slipped by me in a merge to 0.9
-
- This isn't so quick after all, I thought we had a non-SFTP-oriented tilde expansion somewhere, but I'm not seeing one. Looks like that's covered under #9. I've just added a note about `contrib.files.exists` to that ticket so it gets covered.
-
- Applied in changeset commit:e50ec507661a1445d6857242804eeaebc452f071.
-
- Add partial kwarg to contrib.files.append. Fixes #82.
-
- For example, a number of functions were fixed for 0.9.x with backwards compatibility in mind, such as `contrib.files.append`, which added a `partial` kwarg, but defaulted to `True` in order to preserve the old, semi-broken behavior. That kwarg sho...
-
- Applied in changeset commit:5702362eb5e7efa8658a4f66cd89e14258dad68e.
-
- Update changelog to enforce consistent tense
-
- Add docs, make minor name tweak re #141
-
- Docs intro even says these will be there, but they're not. Jerk.
-
- Make cd() overwrite cwd when given an absolute path. Fixes #166.
-
- Applied in changeset commit:2b714ea602e2eff37d8a7d7ec27d45ec6e79353a.
-
- Allow tweaking of more Paramiko params via CLI/env. Implements #141.
-
- Applied in changeset commit:2e4acf932618f9d70bac339c035f93d19f881411.
-
- Re #132, apply to master too.
-
- Fixes #132: strip local() stdout, like run()/sudo() do.
-
- Applied in changeset commit:7f280a0438326d27ab43f0664632a511d040b7e5.
-
- Make context managers restore env upon exception. Fixes #130.
-
- Add changelog entry re #112
-
- Add changelog entry re #95
-
- Applied in changeset commit:4fc541a3122390eb5bb97a2bcf51d65305035c57.
-
- Print fabfile module-level docstring to --list output. Implements #112.
-
- Order changelog entries by ticket number.
- 08:55 PM Bug #110 (Done): Reorder fab CLI arguments list to be alphabetical
- Applied in changeset commit:292f18b63bdc7030760e8032b401c9de8ed0ea4a.
-
- More fixes to alphabetization of fab arg docs Re #110
-
- Add previous commit to changelog
- 08:51 PM Revision 292f18b63bdc7030760e8032b401c9de8ed0ea4a (invoke_shell, 0.9, ...): Fix #110 (alphabetize...
- Fix #110 (alphabetize CLI args document) Used ordering based on long-name. Signed-off-by: Jeff Forcier <jeff@bitprophet.org>
-
- On the one hand, I do hate the broken autodoc. On the other, I don't think a tertiary release is the time to add new dependencies. Bumping this to 1.0.
-
- Applied in changeset commit:4b3b8d7b8d9f637780345a1bbbe0c3d40cee92f4.
-
- Implements #95
-
- Reorg 0.9.1 changelog; add changelog for prev commit
-
- Thanks for the submission and the merge, guys. I was looking for a slightly different behavior when I wrote the ticket, though -- I don't want users to see the no-fabfiles complaint at all, just the help output, in the situation described. Feels c...
-
- Applied in changeset commit:774d6e8d5ee9af635744f76b9875df8327e15ad5.
-
- Print help when no args given and no fabfile. Fixes #75.
-
- Merge branch '0.9' Conflicts: docs/usage/execution.rst fabric/state.py
-
- Partial solution in place, needs more work
-
- One other plus of the shell approach, it seems to obviate the need for a manual echo of my stdin. This is good because it means password prompts work correctly. With the exec_command approach, I have to manually echo stdin, meaning it's impossible...
-
- There's also the problem of it being a real shell session: it displays prompts, which would need to be hidden somehow (set $PS1 to empty string? or what?) and we'd have no way of knowing when the remote end was actually done with the command we're...
-
- Using `Channel.set_combine_stderr(True)` appears to solve the "crossing the streams" problem, though using it long term would mean a total loss of that information (which data is from which stream). ---- As noted in a comment I just made on #80,...
-
- While tinkering with #7 I gave this a shot just for kicks and (at least paired with the select-based stdin/stdout/stderr multiplexing, plus setting `Channel.combine_stderr` to `True`) it works great, outside of the given concerns re: ability to ea...
-
- Trying out more complicated remote programs like `adduser` show that my initial assumption about how stdout and stderr are printed to terminals normally, may be incorrect: I'm getting amusing garbled output by an intermingling of the two streams (...
-
03:42 PM Feature #170: Pull request for exclude_hosts feature
- I've added an option -x / --exclude-hosts which allows you to specify a list of hosts on the command line to exclude from the calculated hosts list (after roles, etc). It also works in the expected way: fab some_func:hosts=a;b,exclude_hosts=a ...
-
- Have it working in the base case now, as evidenced by [this snippet of calling sudo without sudo()](https://gist.github.com/ed7e61fd2b792cf866a9). In other words, the password prompt there is from the remote end (not our usual sudo helper), and I ...
-
- Oh yea, another thing: while I'll probably try to keep it in, the automatic sudo password handling will no longer be necessary once this feature is working robustly, and it's remotely possible I'll drop it and just say "hey, you have full stdin co...
-
- Have been investigating this issue seriously over the last few days since it's started to become a semi-blocker for my use of Fabric at work :) (I am an unabashed itch-scratcher...) Details so far: * I have some basic tests working wherein sys....
-
02:41 AM Support #99: Create subsite for user fabfile/snippet sharing
- What I have used in the past is django-paste (http://code.google.com/p/django-paste/) which is a clone of dpaste.com, not quite as slick as gist but pretty damn close and totally usable. Check out this site http://dpaste.de/. You can then just hav...
05/26/2010
-
- See some thoughts in [this ML thread](http://lists.nongnu.org/archive/html/fab-user/2009-11/msg00014.html). Extracted from it: * Log channel to some persistent storage via a bot * View it on the Web * [irclogger](http://github.com/cschne...
05/25/2010
-
- Review the code (all of 30 seconds worth for the oneliner) :-) and applied this patch cleanly to the latest 0.9 branch. It's FF merge now if ya want from my 0.9-bug75 branch. http://github.com/tswicegood/fabric/tree/0.9-bug75
-
- Have line-by-line input semi working, but worthless without ability to see non-newlined input from remote end
-
- See also [this SO thread](http://stackoverflow.com/questions/760978/long-running-ssh-commands-in-python-paramiko-module-and-how-to-end-them)
-
- would it not make sense to leave this open then? After all the request is for a feature of this redmine install regardless of whether it is implemented in the currently deployed branch or not. And if you upgrade to 0.9 this will be fixed, so WONTF...
-
-
- Redmine 0.8, which we run a modified version of, doesn't seem to have openID support but I'm hoping to upgrade to 0.9 eventually which does seem to have it.
-
- exists on a path with ~ in it does not work, returns 1 even if it exists, because test is called with the path in "%s" so it does not expand. Changing that to simply be %s works but may break something else
-
- please allow logging in to redmine via openid, registration is a small but existing barrier to reporting issues.
Also available in: Atom