Changes by Giampaolo Rodola' g.rod...@gmail.com:
--
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
___
Python-bugs-list mailing list
Senthil Kumaran orsent...@gmail.com added the comment:
On Sun, Apr 11, 2010 at 07:43:56PM +, Éric Araujo wrote:
httplib.Client, ftplib.Client, ftplib.SecureClient would be much more
descriptive than httplib.HTTP and ftplib.FTP. Any interest about adding
aliases?
Aliases would be a bad
Giampaolo Rodola' billiej...@users.sourceforge.net added the comment:
Thinking back about this, I wonder whether FTPS could be a better name to use
instead of FTP_TLS.
It's shorter, easier to remember, and also makes more sense since also SSL can
be used, not only TLS.
--
Éric Araujo mer...@netwok.org added the comment:
It doesn’t look like a constant, too.
httplib.Client, ftplib.Client, ftplib.SecureClient would be much more
descriptive than httplib.HTTP and ftplib.FTP. Any interest about adding aliases?
Regards
--
nosy: +merwok
Antoine Pitrou pit...@free.fr added the comment:
Thinking back about this, I wonder whether FTPS could be a better name to
use instead of FTP_TLS.
It's shorter, easier to remember, and also makes more sense since also SSL
can be used, not only TLS.
What do you mean by also SSL can be
Domen ielect...@gmail.com added the comment:
Nice! Any chance of merging with 2.7? Python3.2 is waaay too far in
future for such useful change to be actually useful.
--
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
Antoine Pitrou pit...@free.fr added the comment:
It's already in 2.7.
--
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
___
Antoine Pitrou pit...@free.fr added the comment:
Here is the current py3k patch I have, after resolving conflicts and
cleaning up the obvious problems.
After tracing a bit, it seems that ssl.wrap_socket() changes the socket
fileno under py3k, while it doesn't under trunk.
--
Added file:
Antoine Pitrou pit...@free.fr added the comment:
Ok, I now have a working patch. The main fix was to change
SSLConnection.secure_connection() to:
def secure_connection(self):
socket = ssl.wrap_socket([ ##etc. ])
self.del_channel()
Changes by Antoine Pitrou pit...@free.fr:
Removed file: http://bugs.python.org/file15350/ftptls-py3k-2.patch
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
Changes by Antoine Pitrou pit...@free.fr:
Added file: http://bugs.python.org/file15351/ftptls-py3k-2.patch
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
Giampaolo Rodola' billiej...@users.sourceforge.net added the comment:
Ok, I took a look and it seems ok to me but I still get some occasional
failures on Windows from time to time.
Because of the threading nature of our server I suspect that moving
del_channel() before ssl.wrap_socket() call,
Antoine Pitrou pit...@free.fr added the comment:
Ok, I took a look and it seems ok to me but I still get some occasional
failures on Windows from time to time.
Because of the threading nature of our server I suspect that moving
del_channel() before ssl.wrap_socket() call, like this:
Ok,
Antoine Pitrou pit...@free.fr added the comment:
Buildbots are ok. Thank you!
--
resolution: accepted - fixed
stage: commit review - committed/rejected
status: open - closed
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
Giampaolo Rodola' billiej...@users.sourceforge.net added the comment:
Can you attach the 3.x patch so that I can test it myself?
I tried to apply the current 2.x patch against the 3.x trunk but I get
conflicts.
--
___
Python tracker
Antoine Pitrou pit...@free.fr added the comment:
Giampaolo, do you plan to add something or is the patch ok to commit?
--
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
Giampaolo Rodola' billiej...@users.sourceforge.net added the comment:
If we want to add SSL support then the patch in attachment modifies the
last one as I described in my previous comment.
I re-run the tests and they are ok so I guess you can go on with the
commit.
--
Added file:
Antoine Pitrou pit...@free.fr added the comment:
The tests don't work under py3k, for some reason I can't figure out.
There's the following error and then the tests hang:
test_acct (test.test_ftplib.TestTLS_FTPClassMixin) ... Exception in
thread Thread-31:
Traceback (most recent call last):
Antoine Pitrou pit...@free.fr added the comment:
I noticed you were using ftp.python.org in the example strings, but that
service doesn't seem to be alive. I don't know if there's another public
FTP-TLS server you could rely on...?
--
___
Python
Domen ielect...@gmail.com added the comment:
What about AUTH SSL? Or is it too-deprecated?
--
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
Giampaolo Rodola' billiej...@users.sourceforge.net added the comment:
You can build the docs by going to the Doc directory and typing make
html there. It isn't critical anyway.
Done. It's well formatted now.
The tests failed to run, I had to replace the KEYCERT declaration with:
KEYCERT
Antoine Pitrou pit...@free.fr added the comment:
The patch is ok to me. Perhaps Bill wants to take a look, otherwise I
think you can commit.
--
resolution: - accepted
stage: patch review - commit review
___
Python tracker rep...@bugs.python.org
Antoine Pitrou pit...@free.fr added the comment:
A last problem:
Traceback (most recent call last):
File stdin, line 1, in module
TypeError: attribute name must be string, not 'classobj'
--
___
Python tracker rep...@bugs.python.org
Giampaolo Rodola' billiej...@users.sourceforge.net added the comment:
A last problem:
Traceback (most recent call last):
File stdin, line 1, in module
TypeError: attribute name must be string, not 'classobj'
Mmmm this doesn't say much.
When does it happen?
Is that the complete traceback
Antoine Pitrou pit...@free.fr added the comment:
Ah, sorry, roundup's e-mail interface ate part of the message.
The error happens when doing from ftplib import *. Apparently __all__
contains a non-string value.
I don't have commit privileges. Someone else should do it.
Ok, I'll do it if Bill
Antoine Pitrou pit...@free.fr added the comment:
Regarding msg64093, the only API change Bill's suggestion would entail
is an additional optional parameter to the constructor, so adding it
later would be backwards-compatible.
--
___
Python tracker
Giampaolo Rodola' billiej...@users.sourceforge.net added the comment:
Ah, sorry, roundup's e-mail interface ate part of the message.
The error happens when doing from ftplib import *. Apparently
__all__
contains a non-string value.
Oh, shame on me! You're right.
Thanks for the great review
Antoine Pitrou pit...@free.fr added the comment:
You can build the docs by going to the Doc directory and typing make
html there. It isn't critical anyway.
The tests failed to run, I had to replace the KEYCERT declaration with:
KEYCERT = os.path.join(os.path.dirname(__file__), keycert.pem)
Giampaolo Rodola' billiej...@users.sourceforge.net added the comment:
A patch including tests and documentation is now in attachment.
The test TLS server is very similar to pyftpdlib's I draw on:
http://code.google.com/p/pyftpdlib/source/browse/trunk/demo/tls_ftpd.py
I wasn't able to compile the
Giampaolo Rodola' billiej...@users.sourceforge.net added the comment:
From a quick look at the patch, if you call login() twice, the socket
will be wrapped twice as well? Perhaps auth_tls() should have a
protection against this.
You're right. Done.
In prot_p() and prot_c(), it seems that
Changes by Giampaolo Rodola' billiej...@users.sourceforge.net:
Removed file: http://bugs.python.org/file15146/ftplib.patch
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
Changes by Giampaolo Rodola' billiej...@users.sourceforge.net:
Added file: http://bugs.python.org/file15147/ftplib.patch
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
Changes by Lance Edgar la...@edbob.org:
--
nosy: +lgedgar
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
___
Python-bugs-list mailing list
Giampaolo Rodola' billiej...@users.sourceforge.net added the comment:
Are there chances to see this reviewed?
--
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
Changes by Antoine Pitrou pit...@free.fr:
Removed file: http://bugs.python.org/file9775/unnamed
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
___
Changes by Antoine Pitrou pit...@free.fr:
Removed file: http://bugs.python.org/file9782/unnamed
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
___
Changes by Antoine Pitrou pit...@free.fr:
Removed file: http://bugs.python.org/file9783/unnamed
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
___
Changes by Antoine Pitrou pit...@free.fr:
Removed file: http://bugs.python.org/file9807/unnamed
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
___
Changes by Antoine Pitrou pit...@free.fr:
Removed file: http://bugs.python.org/file10784/unnamed
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
Changes by Antoine Pitrou pit...@free.fr:
Removed file: http://bugs.python.org/file13161/unnamed
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
Changes by Antoine Pitrou pit...@free.fr:
Removed file: http://bugs.python.org/file13171/unnamed
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
Antoine Pitrou pit...@free.fr added the comment:
Well, first some tests should be added.
As for reviewing, it needs to be done by someone competent with FTP and
SSL/TLS. If no such person is available and you are confident that the
patch is ok (and ready to do necessary maintenance), then
Giampaolo Rodola' billiej...@users.sourceforge.net added the comment:
I can surely write tests altough I think that issue 3890 might cause
some problems since the test server I included some months ago is
asyncore-based and hence asynchronous.
I have a good knowledge of the FTP protocol but
Antoine Pitrou pit...@free.fr added the comment:
From a quick look at the patch, if you call login() twice, the socket
will be wrapped twice as well? Perhaps auth_tls() should have a
protection against this.
In prot_p() and prot_c(), it seems that self._prot_p is updated
unconditionally,
Antoine Pitrou pit...@free.fr added the comment:
One more question, why is ssl_version=ssl.PROTOCOL_TLSv1 hardwired?
--
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
Giampaolo Rodola' billiej...@users.sourceforge.net added the comment:
I released that the attached patch had some indentation issues.
The one in attachment fixes them.
Added file: http://bugs.python.org/file13242/ftplib.patch
___
Python tracker
Giampaolo Rodola' billiej...@users.sourceforge.net added the comment:
I realized that the attached patch had some indentation issues.
The one in attachment fixes them.
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
Changes by Giampaolo Rodola' billiej...@users.sourceforge.net:
Removed file: http://bugs.python.org/file13156/ftplib.patch
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
Jeff Oyama j...@oyama.org added the comment:
Ok after examining it more closely, it appears to be a false alarm, my
apologies
___
Python tracker rep...@bugs.python.org
http://bugs.python.org/issue2054
___
Giampaolo Rodola' billiej...@users.sourceforge.net added the comment:
Actually I have encountered a possible bug. the close()
method doesn't seem to actually close the connection...
Why? What happens exactly?
___
Python tracker rep...@bugs.python.org
Jeff Oyama j...@oyama.org added the comment:
Actually I have encountered a possible bug. the close() method doesn't seem
to actually close the connection...
On Mon, Feb 23, 2009 at 11:56 PM, Jeff Oyama rep...@bugs.python.org wrote:
Jeff Oyama j...@oyama.org added the comment:
Thank you
Giampaolo Rodola' billiej...@users.sourceforge.net added the comment:
After Bill added SSL's unwrap() method I modified my previous patch so
that it shutdown the SSL layer before closing the data connection.
I successfully tested it against proftpd, vsftpd and pyftpdlib TLS
server [1].
If some
Jeff Oyama j...@oyama.org added the comment:
Thank you Giampaolo, it works just as I was hoping, =] I tested it on glftpd
using python 2.6.1.
Added file: http://bugs.python.org/file13161/unnamed
___
Python tracker rep...@bugs.python.org
Jeff Oyama j...@oyama.org added the comment:
Just wondering, has anyone done a patch since Bill made the necessary
changes to ssl.py in order to implement FTP TLS? If so, where can I find
it? I would love to test it out.
--
nosy: +jeffo
___
Python
Changes by Todd Whiteman [EMAIL PROTECTED]:
--
nosy: +twhitema
___
Python tracker [EMAIL PROTECTED]
http://bugs.python.org/issue2054
___
___
Python-bugs-list mailing list
Bill Janssen [EMAIL PROTECTED] added the comment:
I think I'm just going to bring the unwrap already in the _ssl.c code
out to the ssl.py module, that seems to be the simplest fix. Still not
sure you can do a proper fix to ftplib here, but that seems to be a good
thing to do anyway, rather than
Bill Janssen [EMAIL PROTECTED] added the comment:
OK, I think I've done the minimal fix necessary to the SSL module to
allow this work to proceed.
___
Python tracker [EMAIL PROTECTED]
http://bugs.python.org/issue2054
___
Lukasz Szybalski [EMAIL PROTECTED] added the comment:
Is the ftp-tls able to use certificate to connect to ftps server?
I currently need to connect to state's ftps server which requires
certificate to be present when authenticating.
Is that option available? What is the current status of this
Giampaolo Rodola' [EMAIL PROTECTED] added the comment:
Yes, I think that providing an unwrap method for the ssl module would
be good, independently from this issue.
With that implemented and httplib fixed in the way you were mentioning
in this same report I can go on with modifying the ftplib
Bill Janssen [EMAIL PROTECTED] added the comment:
But httplib is far from fixed. It's a nasty tarball of interdependencies...
Bill
On Mon, Jun 30, 2008 at 4:12 AM, Giampaolo Rodola' [EMAIL PROTECTED]
wrote:
Giampaolo Rodola' [EMAIL PROTECTED] added the comment:
Yes, I think that
Giampaolo Rodola' [EMAIL PROTECTED] added the comment:
Could what I've just said be an idea?
___
Python tracker [EMAIL PROTECTED]
http://bugs.python.org/issue2054
___
___
Bill Janssen [EMAIL PROTECTED] added the comment:
The 2.6/3.0 changes are now up-to-date. We could reconsider this
problem. My guess is that we still don't quite know what to do.
I think the issue is that we need a way to unwrap the SSL-secured
TCP stream, after it's been used. So we need to
Giampaolo Rodola' [EMAIL PROTECTED] added the comment:
Bill, are there news about the fix to httplib?
I'd like to see this feature included in 2.6 if possible.
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue2054
__
Bill Janssen [EMAIL PROTECTED] added the comment:
On Fri, Mar 21, 2008 at 5:43 AM, Robert E. [EMAIL PROTECTED] wrote:
Robert E. [EMAIL PROTECTED] added the comment:
Concerning the plain-text login. I think a FTPS class should default to
encrypted login (you could use the ftp class if you
Antoine Pitrou [EMAIL PROTECTED] added the comment:
FWIW, m2crypto already provides an FTP-TLS facility with an
ftplib-compatible API. See http://chandlerproject.org/Projects/MeTooCrypto
--
nosy: +pitrou
__
Tracker [EMAIL PROTECTED]
Giampaolo Rodola' [EMAIL PROTECTED] added the comment:
This is a straightforward implementation of client-side use of SSL,
but it's missing a test case for evaluation. It should include a
patch to test_ftplib to test it.
I'm not sure how it could be tested, since we don't have an FTPS
Bill Janssen [EMAIL PROTECTED] added the comment:
As you point out, the other classes should be fixed. The old client-side
protocol was never very well thought out, IMHO. Continuing to propagate it
would be a mistake.
Bill
On Wed, Mar 19, 2008 at 12:22 PM, Giampaolo Rodola' [EMAIL PROTECTED]
Giampaolo Rodola' [EMAIL PROTECTED] added the comment:
As you point out, the other classes should be fixed. The old
client-side protocol was never very well thought out, IMHO.
Continuing to propagate it would be a mistake.
Ok, how do you think it would have be modified?
Could you provide
Bill Janssen [EMAIL PROTECTED] added the comment:
Probably what I should do is fix httplib, that would provide an example we
could extend to the rest of the modules.
Bill
On Wed, Mar 19, 2008 at 1:46 PM, Giampaolo Rodola' [EMAIL PROTECTED]
wrote:
Giampaolo Rodola' [EMAIL PROTECTED] added the
Bill Janssen [EMAIL PROTECTED] added the comment:
Once I've got JCC working, and finished the SSL work for 2.6.
On Wed, Mar 19, 2008 at 1:46 PM, Giampaolo Rodola' [EMAIL PROTECTED]
wrote:
Giampaolo Rodola' [EMAIL PROTECTED] added the comment:
As you point out, the other classes should be
Domen [EMAIL PROTECTED] added the comment:
The lib should give programmer choice wether to send login through TLS
or not. (as it is described in RFC 4217).
Also, there should be an optional parameter to specify port for ftp
connection.
--
nosy: +iElectric
Bill Janssen added the comment:
This is a straightforward implementation of client-side use of SSL, but
it's missing a test case for evaluation. It should include a patch to
test_ftplib to test it.
Another thing to look at is what the useful arguments are to pass in for
TLS usage over FTP.
Giampaolo Rodola' added the comment:
I've tried to work on this in the last 2 days and here is my
implementation attempt.
The patch in attachment provides a new FTP subclass which connects to
port 21 as usual leaving control and data channels implicitly unprotected.
Securing control and data
New submission from Gregory P. Smith:
ftplib does not support ftp over SSL / TLS as described in RFC 4217.
This would be a nice thing for someone wanting to contribute something
to add.
--
components: Library (Lib)
messages: 62217
nosy: gregory.p.smith
severity: normal
status: open
Changes by Christian Heimes:
--
keywords: +easy
priority: - normal
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue2054
__
___
Python-bugs-list mailing list
75 matches
Mail list logo