GNU bug report logs - #32226
shadowfile test failures

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 32226 in the body.
You can then email your comments to 32226 AT debbugs.gnu.org in the normal way.

Toggle the display of automated, internal messages from the tracker.

Message #3 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Glenn Morris <rgm <at> gnu.org>
To: submit <at> debbugs.gnu.org
Subject: shadowfile test failures
Date: Fri, 20 Jul 2018 13:22:32 -0400

[Message part 1 (text/plain, inline)]

Package: emacs
Version: 26.1.50
Severity: minor

Hi,

Current emacs-26 branch on rhel 7.5, multiple shadowfile test failures,
please see attached log.

[shadowfile-tests.log (application/octet-stream, attachment)]

Message #6 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Glenn Morris <rgm <at> gnu.org>
Cc: 32226 <at> debbugs.gnu.org, michael.albinus <at> gmx.de
Subject: Re: bug#32226: shadowfile test failures
Date: Fri, 20 Jul 2018 22:08:32 +0300

[Message part 1 (text/plain, inline)]

> From: Glenn Morris <rgm <at> gnu.org>
> Date: Fri, 20 Jul 2018 13:22:32 -0400
> Cc: Michael Albinus <michael.albinus <at> gmx.de>
> 
> Current emacs-26 branch on rhel 7.5, multiple shadowfile test failures,
> please see attached log.

On Windows I get failures for tests 06 through 09, log attached.

[shadowfile-tests.log (application/octet-stream, attachment)]

Message #9 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Michael Albinus <michael.albinus <at> gmx.de>
To: Glenn Morris <rgm <at> gnu.org>
Cc: 32226 <at> debbugs.gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Sat, 21 Jul 2018 12:33:42 +0200

Glenn Morris <rgm <at> gnu.org> writes:

> Hi,

Hi Glenn,

> Current emacs-26 branch on rhel 7.5, multiple shadowfile test failures,
> please see attached log.
> 
>       :form
>       (string-equal "my.host.com" "/my.host.com:")
>       :value nil))
>    FAILED   2/10  shadow-test01-sites

Well, I have used "\\w+" as hostname syntax. I've changed this to
"[-.[:word:]]+", could you pls check whether this works?

Pushed to the emacs-26 branch.

Best regards, Michael.

Message #12 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Michael Albinus <michael.albinus <at> gmx.de>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: 32226 <at> debbugs.gnu.org, Glenn Morris <rgm <at> gnu.org>
Subject: Re: bug#32226: shadowfile test failures
Date: Sat, 21 Jul 2018 12:36:57 +0200

Eli Zaretskii <eliz <at> gnu.org> writes:

Hi Eli,

> On Windows I get failures for tests 06 through 09, log attached.

... and tests 00 through 05 are skipped. I've forgotten to add the skip
form in the tests.

Fixed in the emacs-26 branch, could you pls check?

Best regards, Michael.

Message #17 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Michael Albinus <michael.albinus <at> gmx.de>
Cc: 32226 <at> debbugs.gnu.org, rgm <at> gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Sat, 21 Jul 2018 15:03:28 +0300

> From: Michael Albinus <michael.albinus <at> gmx.de>
> Cc: Glenn Morris <rgm <at> gnu.org>,  32226 <at> debbugs.gnu.org
> Date: Sat, 21 Jul 2018 12:36:57 +0200
> 
> > On Windows I get failures for tests 06 through 09, log attached.
> 
> ... and tests 00 through 05 are skipped. I've forgotten to add the skip
> form in the tests.
> 
> Fixed in the emacs-26 branch, could you pls check?

Now all of the tests are skipped, is this expected?

Message #20 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Michael Albinus <michael.albinus <at> gmx.de>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: 32226 <at> debbugs.gnu.org, rgm <at> gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Sat, 21 Jul 2018 14:13:55 +0200

Eli Zaretskii <eliz <at> gnu.org> writes:

Hi Eli,

> Now all of the tests are skipped, is this expected?

Yes, it is for MS Windows. See the definition of
shadow-test-remote-temporary-file-directory.

If you know a temporary directory located on a remote host, you could
run

# env REMOTE_TEMPORARY_FILE_DIRECTORY=/plink:host:/tmp make -C test shadowfile-tests

See the explanation of $REMOTE_TEMPORARY_FILE_DIRECTORY in the
Commentary section of tramp-tests.el. Hmm, maybe I should add a similar
Commentary section to shadowfile-tests.el

Best regards, Michael.

Message #23 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Michael Albinus <michael.albinus <at> gmx.de>
Cc: 32226 <at> debbugs.gnu.org, rgm <at> gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Sat, 21 Jul 2018 15:25:49 +0300

> From: Michael Albinus <michael.albinus <at> gmx.de>
> Cc: rgm <at> gnu.org,  32226 <at> debbugs.gnu.org
> Date: Sat, 21 Jul 2018 14:13:55 +0200
> 
> > Now all of the tests are skipped, is this expected?
> 
> Yes, it is for MS Windows. See the definition of
> shadow-test-remote-temporary-file-directory.
> 
> If you know a temporary directory located on a remote host, you could
> run
> 
> # env REMOTE_TEMPORARY_FILE_DIRECTORY=/plink:host:/tmp make -C test shadowfile-tests
> 
> See the explanation of $REMOTE_TEMPORARY_FILE_DIRECTORY in the
> Commentary section of tramp-tests.el. Hmm, maybe I should add a similar
> Commentary section to shadowfile-tests.el

Hmm, I'm not sure I follow.  For the test purposes, you are actually
shadowing the files to the current system, right?  If so, the location
of the temporary directory is known, right?

Or am I missing something?

Message #26 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Michael Albinus <michael.albinus <at> gmx.de>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: 32226 <at> debbugs.gnu.org, rgm <at> gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Sat, 21 Jul 2018 14:48:38 +0200

Eli Zaretskii <eliz <at> gnu.org> writes:

> Hmm, I'm not sure I follow.  For the test purposes, you are actually
> shadowing the files to the current system, right?  If so, the location
> of the temporary directory is known, right?

I do both. I shadow the files to the local system ("/<system-name>:") and
to a remote system ("/method:host:/dir"). For the latter, I mock-up a
remote temporary directory. This is not possible on MS Windows, because
I use "/bin/sh" as connection program to that virtual remote host,
instead of "ssh" or alike.

So you shall declare a usable remote directory via that environment
variable for the MS WIndows case. If you could give me a recipe how to
setup a mock-up method for MS Windows - the better! It has always beaten
me, that the tests do not run out of the box.

The same mechanism is used for filenotify-tests.el and tramp-tests.el.

Best regards, Michael.

Message #29 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Michael Albinus <michael.albinus <at> gmx.de>
Cc: 32226 <at> debbugs.gnu.org, rgm <at> gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Sat, 21 Jul 2018 16:37:37 +0300

> From: Michael Albinus <michael.albinus <at> gmx.de>
> Cc: rgm <at> gnu.org,  32226 <at> debbugs.gnu.org
> Date: Sat, 21 Jul 2018 14:48:38 +0200
> 
> Eli Zaretskii <eliz <at> gnu.org> writes:
> 
> > Hmm, I'm not sure I follow.  For the test purposes, you are actually
> > shadowing the files to the current system, right?  If so, the location
> > of the temporary directory is known, right?
> 
> I do both. I shadow the files to the local system ("/<system-name>:") and
> to a remote system ("/method:host:/dir"). For the latter, I mock-up a
> remote temporary directory. This is not possible on MS Windows, because
> I use "/bin/sh" as connection program to that virtual remote host,
> instead of "ssh" or alike.

I don't see why this would be impossible.  The test suite is run from
the MSYS Bash anyway, and if I start Emacs from there, I get this:

  (executable-find "sh")
    => "d:/usr/MSYS/bin/sh.exe"

(even though the directory where sh.exe leaves is not on the
system-wide PATH on my systems, although most other users do place it
on PATH).  So use executable-find to get the absolute file name of the
shell, instead of a literal "/bin/sh", and I think everything else
should "just work", no?

> So you shall declare a usable remote directory via that environment
> variable for the MS WIndows case. If you could give me a recipe how to
> setup a mock-up method for MS Windows - the better!

Does the above fit the bill?  If not, what else is needed for the
mock-up method?

Message #32 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: michael.albinus <at> gmx.de
Cc: 32226 <at> debbugs.gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Sat, 21 Jul 2018 16:56:16 +0300

> Date: Sat, 21 Jul 2018 16:37:37 +0300
> From: Eli Zaretskii <eliz <at> gnu.org>
> Cc: 32226 <at> debbugs.gnu.org
> 
>   (executable-find "sh")
>     => "d:/usr/MSYS/bin/sh.exe"
> 
> (even though the directory where sh.exe leaves is not on the
                                          ^^^^^^
"lives", ugh.

Message #35 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Michael Albinus <michael.albinus <at> gmx.de>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: 32226 <at> debbugs.gnu.org, rgm <at> gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Sat, 21 Jul 2018 15:59:21 +0200

Eli Zaretskii <eliz <at> gnu.org> writes:

> I don't see why this would be impossible.  The test suite is run from
> the MSYS Bash anyway, and if I start Emacs from there, I get this:
>
>   (executable-find "sh")
>     => "d:/usr/MSYS/bin/sh.exe"

The test suite is also run interactively (that's what I do), so it
cannot be guaranteed that (executable-find "sh") gives a non-nil
result. For this case, there is the REMOTE_TEMPORARY_FILE_DIRECTORY trick.

> (even though the directory where sh.exe leaves is not on the
> system-wide PATH on my systems, although most other users do place it
> on PATH).  So use executable-find to get the absolute file name of the
> shell, instead of a literal "/bin/sh", and I think everything else
> should "just work", no?
>
>> So you shall declare a usable remote directory via that environment
>> variable for the MS WIndows case. If you could give me a recipe how to
>> setup a mock-up method for MS Windows - the better!
>
> Does the above fit the bill?  If not, what else is needed for the
> mock-up method?

I will check your proposal next week, when I'm at work. Maybe it works,
but likely I will find only an MS Windows machine which does not offer
any sh.exe - that's Murphy. In that case, I will produce a patch which
you could check.

(All of this shouldn't stop us from closing this bug report, after
confirmation of Glenn)

Thanks, and best regards, Michael.

Message #38 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Michael Albinus <michael.albinus <at> gmx.de>
Cc: 32226 <at> debbugs.gnu.org, rgm <at> gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Sat, 21 Jul 2018 17:21:58 +0300

> From: Michael Albinus <michael.albinus <at> gmx.de>
> Cc: rgm <at> gnu.org,  32226 <at> debbugs.gnu.org
> Date: Sat, 21 Jul 2018 15:59:21 +0200
> 
> > I don't see why this would be impossible.  The test suite is run from
> > the MSYS Bash anyway, and if I start Emacs from there, I get this:
> >
> >   (executable-find "sh")
> >     => "d:/usr/MSYS/bin/sh.exe"
> 
> The test suite is also run interactively (that's what I do), so it
> cannot be guaranteed that (executable-find "sh") gives a non-nil
> result. For this case, there is the REMOTE_TEMPORARY_FILE_DIRECTORY trick.

We could instead ask people to set explicit-shell-file-name.

> (All of this shouldn't stop us from closing this bug report, after
> confirmation of Glenn)

Right.

Message #41 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Michael Albinus <michael.albinus <at> gmx.de>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: 32226 <at> debbugs.gnu.org, rgm <at> gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Sat, 21 Jul 2018 16:41:15 +0200

Eli Zaretskii <eliz <at> gnu.org> writes:

>> The test suite is also run interactively (that's what I do), so it
>> cannot be guaranteed that (executable-find "sh") gives a non-nil
>> result. For this case, there is the REMOTE_TEMPORARY_FILE_DIRECTORY trick.
>
> We could instead ask people to set explicit-shell-file-name.

That's not related directly. REMOTE_TEMPORARY_FILE_DIRECTORY shall be a,
hmm, temporary remote directory. It is useful for testing many
configurations. Before I release a new Tramp version, I run stress tests
for several dozen targets, covering all different connection methods and
configuration winkles, and also different Emacs versions starting with
Emacs 24.

Such a campaign (a script) runs usually longer than 24h, and it never
passes all tests during the first run ...

Using a local sh is interesting only for the default mock-up
method. This is the most boring constellation in my own tests, because
everybody else (and hydra) run those tests already.

Best regards, Michael.

Message #44 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Michael Albinus <michael.albinus <at> gmx.de>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: 32226 <at> debbugs.gnu.org, rgm <at> gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Mon, 23 Jul 2018 11:55:09 +0200

[Message part 1 (text/plain, inline)]

Eli Zaretskii <eliz <at> gnu.org> writes:

Hi Eli,

> Does the above fit the bill?  If not, what else is needed for the
> mock-up method?

I have tried the following patch. It still runs as expected on my
GNU/Linux machine. Could you pls check for MS Windows?

If it works there, I'll commit to the emacs-26 branch (and similar
patches for filenotify-tests.el and tramp-tests.el).

[Message part 2 (text/plain, attachment)]

Message #47 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Michael Albinus <michael.albinus <at> gmx.de>
Cc: 32226 <at> debbugs.gnu.org, rgm <at> gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Mon, 23 Jul 2018 18:59:32 +0300

[Message part 1 (text/plain, inline)]

> From: Michael Albinus <michael.albinus <at> gmx.de>
> Cc: rgm <at> gnu.org,  32226 <at> debbugs.gnu.org
> Date: Mon, 23 Jul 2018 11:55:09 +0200
> 
> I have tried the following patch. It still runs as expected on my
> GNU/Linux machine. Could you pls check for MS Windows?

I ran that in batch mode.  It ran for a while, and then seemed to
hang.  I interrupted it; what ended up in the log file is attached.

Let me know how can I assist you in debugging this.

[shadowfile-tests.log (application/octet-stream, attachment)]

Message #50 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Michael Albinus <michael.albinus <at> gmx.de>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: 32226 <at> debbugs.gnu.org, rgm <at> gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Mon, 23 Jul 2018 18:18:14 +0200

Eli Zaretskii <eliz <at> gnu.org> writes:

Hi Eli,

> I ran that in batch mode.  It ran for a while, and then seemed to
> hang.  I interrupted it; what ended up in the log file is attached.
>
> Let me know how can I assist you in debugging this.
>
> Running 10 tests (2018-07-23 18:52:28+0300)
>    passed   1/10  shadow-test00-clusters

Well, at least the (executable-find "sh") trick does it. The tests
run. So I will apply it to the three test files I've mentioned
earlier. Emacs-26 or master branch?

>       (equal
>        (tramp-file-name "cluster" nil nil "c" nil "/DOCUME~1/Zaretzky/LOCALS~1/Temp/shadowfile-testsSe3Hgz" nil)
>        (tramp-file-name nil nil nil "HOME-C4E4A596F7" nil "c:/DOCUME~1/Zaretzky/LOCALS~1/Temp/shadowfile-testsSe3Hgz" nil))

Obviously an error I need to fix. I'm wondering, whether shadowfile.el
has run ever on MS Windows. I bed, the same error would happen with the
old implementation.

Thanks for testing.

Best regards, Michael.

Message #53 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Michael Albinus <michael.albinus <at> gmx.de>
Cc: 32226 <at> debbugs.gnu.org, rgm <at> gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Mon, 23 Jul 2018 19:39:49 +0300

> From: Michael Albinus <michael.albinus <at> gmx.de>
> Cc: rgm <at> gnu.org,  32226 <at> debbugs.gnu.org
> Date: Mon, 23 Jul 2018 18:18:14 +0200
> 
> > Running 10 tests (2018-07-23 18:52:28+0300)
> >    passed   1/10  shadow-test00-clusters
> 
> Well, at least the (executable-find "sh") trick does it. The tests
> run. So I will apply it to the three test files I've mentioned
> earlier. Emacs-26 or master branch?

Emacs-26 is okay in this case, since we are fixing something that is
broken already.

> I'm wondering, whether shadowfile.el has run ever on MS Windows.

I wouldn't count on it.

Thanks.

Message #56 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Michael Albinus <michael.albinus <at> gmx.de>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: 32226 <at> debbugs.gnu.org, rgm <at> gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Mon, 23 Jul 2018 18:52:06 +0200

Eli Zaretskii <eliz <at> gnu.org> writes:

>> Well, at least the (executable-find "sh") trick does it. The tests
>> run. So I will apply it to the three test files I've mentioned
>> earlier. Emacs-26 or master branch?
>
> Emacs-26 is okay in this case, since we are fixing something that is
> broken already.

Will do. Rather tomorrow, I'll have the chance to test on MS
Windows. Might happen that the other two test files uncover further
errors on MS Windows ...

>> I'm wondering, whether shadowfile.el has run ever on MS Windows.
>
> I wouldn't count on it.

I don't :-)

> Thanks.

Best regards, Michael.

Message #59 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Glenn Morris <rgm <at> gnu.org>
To: Michael Albinus <michael.albinus <at> gmx.de>
Cc: 32226 <at> debbugs.gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Tue, 24 Jul 2018 13:44:51 -0400

Michael Albinus wrote:

> Well, I have used "\\w+" as hostname syntax. I've changed this to
> "[-.[:word:]]+", could you pls check whether this works?
>
> Pushed to the emacs-26 branch.

Works for me thanks. Separately, please note that
shadow-test08-shadow-todo has aborted on hydra.nixos.org ever since it
was first added.

Message #62 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Michael Albinus <michael.albinus <at> gmx.de>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: 32226 <at> debbugs.gnu.org, rgm <at> gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Wed, 25 Jul 2018 12:48:50 +0200

Eli Zaretskii <eliz <at> gnu.org> writes:

Hi Eli,

> I ran that in batch mode.  It ran for a while, and then seemed to
> hang.  I interrupted it; what ended up in the log file is attached.

I could reproduce it with Emacs 26 on MS Windows.

> Let me know how can I assist you in debugging this.

>       (equal
>        (tramp-file-name "cluster" nil nil "c" nil "/DOCUME~1/Zaretzky/LOCALS~1/Temp/shadowfile-testsSe3Hgz" nil)
>        (tramp-file-name nil nil nil "HOME-C4E4A596F7" nil "c:/DOCUME~1/Zaretzky/LOCALS~1/Temp/shadowfile-testsSe3Hgz" nil))

Well, the problem is the syntax of local file names on MS Windows, like
"c:/DOCUME~1/Zaretzky/LOCALS~1/Temp/shadowfile-testsSe3Hgz". shadowfile
uses cluster file names with syntax "/<cluster name>:<local file name>".
If the <local file name> starts with a volume letter, the resulting
string looks like a remote file name, and shadowfile is confused.

Fixing this is not trivial. Is it worth to do it? Again, I doubt that
shadowfile has worked ever on MS Windows; I believe it would be
sufficient to document this restriction.

Best regards, Michael.

Message #65 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Michael Albinus <michael.albinus <at> gmx.de>
To: Glenn Morris <rgm <at> gnu.org>
Cc: 32226 <at> debbugs.gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Wed, 25 Jul 2018 13:22:03 +0200

Glenn Morris <rgm <at> gnu.org> writes:

> Separately, please note that shadow-test08-shadow-todo has aborted on
> hydra.nixos.org ever since it was first added.

No further information, hmm. I've made a small compatibility change in
shadowfile.el, which might fix it, or not. And I've instrumented
shadow-test08-shadow-todo to be more chatty. Let's see.

Best regards, Michael.

Message #68 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Michael Albinus <michael.albinus <at> gmx.de>
Cc: 32226 <at> debbugs.gnu.org, rgm <at> gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Wed, 25 Jul 2018 17:47:09 +0300

> From: Michael Albinus <michael.albinus <at> gmx.de>
> Cc: rgm <at> gnu.org,  32226 <at> debbugs.gnu.org
> Date: Wed, 25 Jul 2018 12:48:50 +0200
> 
> >       (equal
> >        (tramp-file-name "cluster" nil nil "c" nil "/DOCUME~1/Zaretzky/LOCALS~1/Temp/shadowfile-testsSe3Hgz" nil)
> >        (tramp-file-name nil nil nil "HOME-C4E4A596F7" nil "c:/DOCUME~1/Zaretzky/LOCALS~1/Temp/shadowfile-testsSe3Hgz" nil))
> 
> Well, the problem is the syntax of local file names on MS Windows, like
> "c:/DOCUME~1/Zaretzky/LOCALS~1/Temp/shadowfile-testsSe3Hgz". shadowfile
> uses cluster file names with syntax "/<cluster name>:<local file name>".
> If the <local file name> starts with a volume letter, the resulting
> string looks like a remote file name, and shadowfile is confused.

Can convert-standard-filename help?  It will convert any colon except
the first one to '!'.

> Fixing this is not trivial. Is it worth to do it? Again, I doubt that
> shadowfile has worked ever on MS Windows; I believe it would be
> sufficient to document this restriction.

It's up to you.  If fixing is indeed hard, I'd hate to get in the way
of your development of threaded Tramp ;-)

Message #71 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Michael Albinus <michael.albinus <at> gmx.de>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: 32226 <at> debbugs.gnu.org, rgm <at> gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Wed, 25 Jul 2018 17:19:55 +0200

Eli Zaretskii <eliz <at> gnu.org> writes:

Hi Eli,

>> Well, the problem is the syntax of local file names on MS Windows, like
>> "c:/DOCUME~1/Zaretzky/LOCALS~1/Temp/shadowfile-testsSe3Hgz". shadowfile
>> uses cluster file names with syntax "/<cluster name>:<local file name>".
>> If the <local file name> starts with a volume letter, the resulting
>> string looks like a remote file name, and shadowfile is confused.
>
> Can convert-standard-filename help?  It will convert any colon except
> the first one to '!'.

That was also my first idea. But it doesn't help:
"/myname:c:/DOCUME~1/Zaretzky/LOCALS~1/Temp/shadowfile-testsSe3Hgz" is
regarded as remote file name due to the colon after the drive letter. I
would need to change the notation of cluster file names, something
different from "/name:local name". This would introduce much more
incompatibility for all other shadoefile users. And again, I'm convinced
there hasn't been a shadowfile user on MS Windows ever.

>> Fixing this is not trivial. Is it worth to do it? Again, I doubt that
>> shadowfile has worked ever on MS Windows; I believe it would be
>> sufficient to document this restriction.
>
> It's up to you.  If fixing is indeed hard, I'd hate to get in the way
> of your development of threaded Tramp ;-)

I guess best would be to document this restriction. We can still react,
when furious bug reports arrive us. Which I doubt.

I will add (skip-unless (not (memq system-type '(windows-nt ms-dos))))
to all shadowfile-tests tests in the emacs-26 branch.

Message #76 received at 32226-done <at> debbugs.gnu.org (full text, mbox):

From: Michael Albinus <michael.albinus <at> gmx.de>
To: Glenn Morris <rgm <at> gnu.org>
Cc: 32226-done <at> debbugs.gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Sun, 12 Aug 2018 18:19:11 +0200

Version: 26.2

> Glenn Morris <rgm <at> gnu.org> writes:
>
>> Separately, please note that shadow-test08-shadow-todo has aborted on
>> hydra.nixos.org ever since it was first added.
>
> No further information, hmm. I've made a small compatibility change in
> shadowfile.el, which might fix it, or not. And I've instrumented
> shadow-test08-shadow-todo to be more chatty. Let's see.

After weeks of hunting, I could trap it finally. Fixed in master, all
shadowfile-tests tests pass successfully on hydra.

Since I've backported the fix to the emacs-26 branch, I'm closing this
bug for Emacs 26.2.

Best regards, Michael.

Message #79 received at 32226 <at> debbugs.gnu.org (full text, mbox):

From: Glenn Morris <rgm <at> gnu.org>
To: Michael Albinus <michael.albinus <at> gmx.de>
Cc: 32226 <at> debbugs.gnu.org
Subject: Re: bug#32226: shadowfile test failures
Date: Sun, 12 Aug 2018 19:46:19 -0400

Michael Albinus wrote:

> After weeks of hunting, I could trap it finally. Fixed in master, all
> shadowfile-tests tests pass successfully on hydra.

Quite a saga, well done! :)

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.