New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
TypeError: Expected unicode or bytes, got <read-only buffer for 0x10fb7e120, size 5242880, offset 0 at 0x10fb79eb0> #285
Comments
N.B. full traceback was posted in the above links (is required to figure out what's going on...) Replicating the useful parts here:
The relevant part of The processing via |
All that aside, @basictheprogram you should be able to skip the Paramiko side of this bug by downgrading your local Paramiko to 1.12 or older - only 1.13 adds this breaking change. If you don't care about Python 3 support (which seems likely given the other codebases' reliance on To get this fixed for 1.13+, I'd need a reproduction method - I don't run any bzr repositories so I'd need a quick walkthrough of how to reproduce this. (Ideally it would be in a test suite somewhere and I could then just run that test...?) |
@bitprophet You can reproduce this with
On a Debian based system, you'd need to install |
It looks like PR #352 by jelmer has been submitted as a solution have we validated it fixes it? |
Has anyone else tested the #352 PR as an appropriate fix for this issue? I can't test the Debian package as the bzr package maintainer has seen fit to hard code a Conflict against the paramiko package so being able to install both packages and run the test suite is being made harder for me to validate. |
I can confirm it the
|
This is currently causing the ansible package to conflict with bzr on Debian, and will potentially cause the removal of python-paramiko (and maybe bzr) from Debian, along with everything that depends on it. With the freeze for Jessie coming up soon, the window is rapidly narrowing for this issue to get fixed in time. |
On Sun, Aug 24, 2014 at 05:27:55PM -0700, Jeremy T. Bouse wrote:
You should be able to reproduce the error and the fix using the bzr and then running the tests from the unpacked tarball ./bzr selftest --no-plugins (Or by just building the bzr package with the Conflicts dropped) Cheers, Jelmer |
@hlieberman Even if I (the python-paramiko maintainer) get this patch included and uploaded in a fixed version bzr will still conflict because the bzr maintainer team decided to put a "Conflicts: python-paramiko (>> 1.12.0)" into the debian/control which doesn't help matters just exacerbates the problem. I'm sitting here in the DC14 Hacklab working on packaging issues and have discussed that move with several other DDs who think it was a poor move on the part of the bzr team. |
@jbouse I totally understand why you're unhappy with the situation, adding a Conflicts certainly wasn't anyone's prefered solution. It was done to prevent bzr's removal from testing after it became clear that jelmer's attempts to get this fix in would not be timely enough. I'm also at DebConf (I'm asb@d.o) and I'm an uploader of the bzr package. I'd be quite happy to coordinate an upload of bzr that removes the Conflicts. Anyways... Let's move this off the upstream issue tracker. I'd be happy to discuss this with you over a drink of some kind here in Portland. |
(CC: debian bugs) On Mon, Aug 25, 2014 at 08:50:17AM -0700, Jeremy T. Bouse wrote:
This is a regression in paramiko, and this fix is a trivial and That would have allowed the bug to be fixed in a timely manner % apt-cache rdepends bzr python-bzrlib | grep -v "|" | uniq | wc -l A Conflict was the quickest way in which I could unbreak bzr for its An alternative workaround for me is to patch bzr to disable its paramiko If there's anything further I can do to help get this bug fixed in paramiko Jelmer Jelmer Vernooij jelmer@samba.org - https://jelmer.uk/ |
On Mon, Aug 25, 2014 at 09:06:16AM -0700, Andrew SB wrote:
Thanks Andrew, I think this bug could do with some higher-bandwidth Jelmer Vernooij jelmer@samba.org - https://jelmer.uk/ |
Hi, maintainer here. Last time I looked at this the severity re: packaging wasn't nearly as clear - that's super unfortunate :( I'll do what I can to get a bugfix release out today or tomorrow. Thanks for all the input! |
Confirmed replication of error as follows (on a Mac OS X 10.8 system):
Now testing out the linked fix... |
Yup, works fine. Regular test suite still passes too. Changelogging & merging, will close this post-feedback from the affected users. |
Merged and pushed to master. Backporting to release branches momentarily. |
Conflicts: sites/www/changelog.rst
This is now in the 1.13 and 1.14 lines in addition to master. I'll be pushing out a release tomorrow sometime, probably the early half of the day Pacific time. Again, feedback before then is welcome. |
1.13.2 and 1.14.1 are on PyPI now. (This isn't what I meant by "the early half of the day tomorrow", but hey, if the shoe fits.) |
Thank you @bitprophet. I've found the 1.14.1 release tarball. I'm just having some local package maintenance repo issues of my own fault but I should have the Debian package produced and uploaded soon which should make @jelmer very happy. |
Glad to hear it @jbouse, please keep me posted, and thanks :) |
…ted. This fixes bzr's use of paramiko. Fixes issue paramiko#343/paramiko#285.
Trying to get bzr push sftp:// working.
bzr with remote repositories requires paramiko and pycrypto.
pycrypto won't compile on clang-5.1 without a patch.
The bzr bug report is here
https://bugs.launchpad.net/bzr/+bug/1293257
The pycrypto bug report is here
https://bugs.launchpad.net/bzr/+bug/1293257
The exception is thrown from paramiko
But the only thing that changed with the patch to pycrypto to get it to build on clang-5.1. Just attempting to get all developers of all the components on the same page.
The text was updated successfully, but these errors were encountered: