Re: [fossil-users] strange problem with graphical diff

[j. van den hoff]

On Wed, 23 Aug 2017 22:28:03 +0200, Richard Hipp  wrote:


On 8/23/17, j. van den hoff  wrote:

On Wed, 23 Aug 2017 19:54:52 +0200, Richard Hipp  wrote:


unable to create directory /var


What happens when you try running this command on the latest trunk
check-in?


works again. no more hiccups. problem solved it seems. thanks a lot for
looking into this.



The problem is not yet completely solved.  It happened to fix your
case because the symbolic link is toward the beginning of the pathname
and the code now only checks the beginning of the path if the whole
path does not yet exist.

Removing support for --no-dir-symlinks will fix the problem.  But I
need to wait for feedback before backing out that feature.


understood. thanks.


--
Using Opera's revolutionary email client: http://www.opera.com/mail/
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] strange problem with graphical diff

[Richard Hipp]

On 8/23/17, j. van den hoff  wrote:
> On Wed, 23 Aug 2017 19:54:52 +0200, Richard Hipp  wrote:
>
>>> unable to create directory /var
>>
>> What happens when you try running this command on the latest trunk
>> check-in?
>
> works again. no more hiccups. problem solved it seems. thanks a lot for
> looking into this.
>

The problem is not yet completely solved.  It happened to fix your
case because the symbolic link is toward the beginning of the pathname
and the code now only checks the beginning of the path if the whole
path does not yet exist.

Removing support for --no-dir-symlinks will fix the problem.  But I
need to wait for feedback before backing out that feature.
-- 
D. Richard Hipp
d...@sqlite.org
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] strange problem with graphical diff

[j. van den hoff]

On Wed, 23 Aug 2017 19:54:52 +0200, Richard Hipp  wrote:


unable to create directory /var


What happens when you try running this command on the latest trunk  
check-in?


works again. no more hiccups. problem solved it seems. thanks a lot for  
looking into this.







--
Using Opera's revolutionary email client: http://www.opera.com/mail/
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] strange problem with graphical diff

[Richard Hipp]

> unable to create directory /var

What happens when you try running this command on the latest trunk check-in?

-- 
D. Richard Hipp
d...@sqlite.org
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] strange problem with graphical diff

[Eric Dillon]

On Wed, Aug 23, 2017, 12:19 PM Richard Hipp  wrote:

And it should now be fixed on trunk.
>

Verified. Thanks.
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] strange problem with graphical diff

[Richard Hipp]

On 8/23/17, Richard Hipp  wrote:
> On 8/23/17, Eric Dillon  wrote:
>> Here is the error generated in Windows. Works correctly without the --to
>> parameter.
>>
>> 10:19:38 $ fo gdiff --from d5b015 --to 810dd0
>>
>> Index: src/http_transport.c
>> ==
>> the
>> [C:\Users\xx\AppData\Local\Temp\/src/http_transport.c-v1-6pCA3817v8hKvMt.txt]
>> argument to the "C:\BIN\WinMerge-2.14.0-exe\winmergeu.exe" command
>> contains
>> a character (ascii 0x5c) that is a security risk
>
> This is an unrelated problem.

And it should now be fixed on trunk.

-- 
D. Richard Hipp
d...@sqlite.org
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] strange problem with graphical diff

[j. van den hoff]

On Wed, 23 Aug 2017 18:09:56 +0200, Richard Hipp  wrote:

more important: is a fix/work-around possible (apart from telling me to  
do

it myself which I would have a hard time with ...)



Hold on there, honcho.  We are not your employees. This is a community
effort.  Expressing impatience with the debugging process is not an
effective strategy for recruiting volunteer helpers to work on the
problem.


I beg your pardon? you've got that completely wrong... by no means was  
this intended to signal/express impatience on my side. I am *not*  
impatient. and no need to explain the very obvious to me (regarding  
community effort and you not being my employees...), really. I appreciate  
the work invested into fossil as much as the next guy.


I was just asking whether a fix (or some user side work around I am not  
aware of) is possible/available. nothing more, ok? and the parenthesis  
just indicated that possibly proposing I contribute a patch would not be  
realistic due to lack of expertise on my side in C programming... that's  
all.  hope this is straightened out now..



--
Using Opera's revolutionary email client: http://www.opera.com/mail/
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] strange problem with graphical diff

[Richard Hipp]

On 8/23/17, Eric Dillon  wrote:
> Here is the error generated in Windows. Works correctly without the --to
> parameter.
>
> 10:19:38 $ fo gdiff --from d5b015 --to 810dd0
>
> Index: src/http_transport.c
> ==
> the
> [C:\Users\xx\AppData\Local\Temp\/src/http_transport.c-v1-6pCA3817v8hKvMt.txt]
> argument to the "C:\BIN\WinMerge-2.14.0-exe\winmergeu.exe" command contains
> a character (ascii 0x5c) that is a security risk

This is an unrelated problem.

-- 
D. Richard Hipp
d...@sqlite.org
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] strange problem with graphical diff

[Eric Dillon]

Here's a bisect of the error on Windows.

11:39:32 $ fo bisect good
bisect complete
  2 BAD 2017-08-21 12:18:49 810dd031ec3ee1d6
  5 BAD 2017-08-12 18:24:29 9eea719af68932ad
  6 BAD 2017-08-12 18:20:26 9690d370e0714a43
  7 BAD 2017-08-12 18:15:36 3b191c984b831571
  8 GOOD2017-08-12 04:19:56 45a3d4b1670f123e CURRENT
  4 GOOD2017-08-11 16:00:55 3ebbe7bcaa8f2221
  3 GOOD2017-07-28 18:41:13 04de083ec842c3f2
  1 GOOD2017-07-10 18:19:00 b1a7527b73a906a0


On Wed, Aug 23, 2017, 11:20 AM Richard Hipp  wrote:

> On 8/23/17, j. van den hoff  wrote:
> > unable to create directory /var
>
> You might be able to work around the problem by adding the
> --no-dir-symlinks command-line option.
> --
> D. Richard Hipp
> d...@sqlite.org
> ___
> fossil-users mailing list
> fossil-users@lists.fossil-scm.org
> http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
>
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] strange problem with graphical diff

[Richard Hipp]

On 8/23/17, j. van den hoff  wrote:
> unable to create directory /var

You might be able to work around the problem by adding the
--no-dir-symlinks command-line option.
-- 
D. Richard Hipp
d...@sqlite.org
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] strange problem with graphical diff

[Richard Hipp]

On 8/23/17, j. van den hoff  wrote:
>
> I reiterate that I wonder what has changed recently (either on the side of
> OSX or fossil) since the problem was definitely non-existent some time
> ago?

The core problem is that sometimes it is desirable to follow symbolic
links and sometimes it is not.  And it is not always obvious which
choice to make.  Often the choice depends on context which is only
available way up in the call stack and which is not passed down into
the routines that are actually making the choice.

Support for symbolic links is problematic.  It has always been so.
This is true for all software, not just Fossil.

.
>
> more important: is a fix/work-around possible (apart from telling me to do
> it myself which I would have a hard time with ...)

Hold on there, honcho.  We are not your employees. This is a community
effort.  Expressing impatience with the debugging process is not an
effective strategy for recruiting volunteer helpers to work on the
problem.

-- 
D. Richard Hipp
d...@sqlite.org
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] strange problem with graphical diff

[Eric Dillon]

Here is the error generated in Windows. Works correctly without the --to
parameter.

10:19:38 $ fo gdiff --from d5b015 --to 810dd0

Index: src/http_transport.c
==
the
[C:\Users\xx\AppData\Local\Temp\/src/http_transport.c-v1-6pCA3817v8hKvMt.txt]
argument to the "C:\BIN\WinMerge-2.14.0-exe\winmergeu.exe" command contains
a character (ascii 0x5c) that is a security risk

On Wed, Aug 23, 2017, 10:24 AM j. van den hoff  wrote:

> On Wed, 23 Aug 2017 17:14:05 +0200, Warren Young 
> wrote:
>
> > On Aug 23, 2017, at 7:21 AM, Richard Hipp  wrote:
> >>
> >> On 8/23/17, j. van den hoff  wrote:
> >>> unable to create directory /var
> >>
> >> It is trying to create a temporary file in which to store the one of
> >> the two sides of the diff.  Can you trace the problem by running in a
> >> debugger?
> >
> > This sounds like a repeat of:
> >
> >
> https://www.mail-archive.com/fossil-users@lists.fossil-scm.org/msg24920.html
> >
> > Basically, it’s an artifact of the way macOS symlinks /var, /tmp, etc.
> > to their actual locations, which doesn’t happen on other *ix type
> > systems, so the bug is never tickled there.
>
> yes, this looks like the exact same problem, see my previous mail: fossil
> detects that `/var' exists while not being a directory and bails out
> (which it should not but rather resolve the link and check whether the
> resolved link is a directory...).
>
> I reiterate that I wonder what has changed recently (either on the side of
> OSX or fossil) since the problem was definitely non-existent some time
> ago? and why does it work when omitting `--to' from the `gdiff' call? no
> idea...
>
> more important: is a fix/work-around possible (apart from telling me to do
> it myself which I would have a hard time with ...)
>
>
> > ___
> > fossil-users mailing list
> > fossil-users@lists.fossil-scm.org
> > http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
>
>
> --
> Using Opera's revolutionary email client: http://www.opera.com/mail/
> ___
> fossil-users mailing list
> fossil-users@lists.fossil-scm.org
> http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
>
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] strange problem with graphical diff

[j. van den hoff]

On Wed, 23 Aug 2017 17:14:05 +0200, Warren Young   
wrote:



On Aug 23, 2017, at 7:21 AM, Richard Hipp  wrote:


On 8/23/17, j. van den hoff  wrote:

unable to create directory /var


It is trying to create a temporary file in which to store the one of
the two sides of the diff.  Can you trace the problem by running in a
debugger?


This sounds like a repeat of:

https://www.mail-archive.com/fossil-users@lists.fossil-scm.org/msg24920.html

Basically, it’s an artifact of the way macOS symlinks /var, /tmp, etc.  
to their actual locations, which doesn’t happen on other *ix type  
systems, so the bug is never tickled there.


yes, this looks like the exact same problem, see my previous mail: fossil  
detects that `/var' exists while not being a directory and bails out  
(which it should not but rather resolve the link and check whether the  
resolved link is a directory...).


I reiterate that I wonder what has changed recently (either on the side of  
OSX or fossil) since the problem was definitely non-existent some time  
ago? and why does it work when omitting `--to' from the `gdiff' call? no  
idea...


more important: is a fix/work-around possible (apart from telling me to do  
it myself which I would have a hard time with ...)




___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users



--
Using Opera's revolutionary email client: http://www.opera.com/mail/
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] strange problem with graphical diff

[Richard Hipp]

The problem originates at this check-in:
https://fossil-scm.org/fossil/timeline?c=2375d6cbce933267

-- 
D. Richard Hipp
d...@sqlite.org
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] strange problem with graphical diff

[Warren Young]

On Aug 23, 2017, at 7:21 AM, Richard Hipp  wrote:
> 
> On 8/23/17, j. van den hoff  wrote:
>> unable to create directory /var
> 
> It is trying to create a temporary file in which to store the one of
> the two sides of the diff.  Can you trace the problem by running in a
> debugger?

This sounds like a repeat of:

https://www.mail-archive.com/fossil-users@lists.fossil-scm.org/msg24920.html

Basically, it’s an artifact of the way macOS symlinks /var, /tmp, etc. to their 
actual locations, which doesn’t happen on other *ix type systems, so the bug is 
never tickled there.
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] strange problem with graphical diff

[j. van den hoff]

On Wed, 23 Aug 2017 15:21:46 +0200, Richard Hipp  wrote:


On 8/23/17, j. van den hoff  wrote:

unable to create directory /var



It is trying to create a temporary file in which to store the one of
the two sides of the diff.  Can you trace the problem by running in a


and it does that differently, depending on whether the `--to' option is  
explicitly specified (as indicated it does not work with the `--to' even  
if the flag is redundant and pointing to the CURRENT revision...)?



debugger?


not easily (OSX, clang rather than gcc etc.: don't even know whether gdb  
would work or which other debugger to use ...). so I have tried to trace  
it down the pedestrian way. result:


the error message is the one in line 650 of `file.c':

648 if( file_mkdir(zName, forceFlag) &&  
file_wd_isdir(zName)!=1 ){

649   if (errorReturn <= 0) {
650 fossil_fatal_recursive("unable to create directory  
%s", zName);

651   }
652   rc = errorReturn;
653   break;
654 }


file_mkdir returns -1, wd_isdir returns 2 and errno is 17 (file exists).  
this is of course true: under OSX `/var' is a link to `/private/var': so  
the file `/var' exists and it is not a directory (but rather is a link and  
points to one). so it seems that the test in `file.c' would have to  
account for the possibility that `/var' is a soft link pointing to some  
directory, no? simply skipping the whole block makes `gdiff' work again...


what I do not quite understand: I am sure that `gdiff' worked fine  
previously and that `/var' "always" was a soft link. so what has changed  
in `fossil' recently causing this (seeming) regression?


if I can provide further information, please let me know.

thank you

joerg








--
Using Opera's revolutionary email client: http://www.opera.com/mail/
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] strange problem with graphical diff

[Andy Goth]

I would also suggest finding an old version of Fossil that doesn't exhibit
the problem and using bisect to pinpoint when the behavior changed.

On Wed, Aug 23, 2017 at 8:21 AM, Richard Hipp  wrote:

> On 8/23/17, j. van den hoff  wrote:
> > unable to create directory /var
> >
>
> It is trying to create a temporary file in which to store the one of
> the two sides of the diff.  Can you trace the problem by running in a
> debugger?
>
> --
> D. Richard Hipp
> d...@sqlite.org
> ___
> fossil-users mailing list
> fossil-users@lists.fossil-scm.org
> http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
>



-- 
Andy Goth | 
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] strange problem with graphical diff

[Richard Hipp]

On 8/23/17, j. van den hoff  wrote:
> unable to create directory /var
>

It is trying to create a temporary file in which to store the one of
the two sides of the diff.  Can you trace the problem by running in a
debugger?

-- 
D. Richard Hipp
d...@sqlite.org
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

[fossil-users] strange problem with graphical diff

[j. van den hoff]

today I have stumbled over this problem:

issuing something like

fossil gdiff --from 823c95ff8a --to eac7dff4fe

yields the terminal output (in this example):

8<-

Index: Makefile
==
unable to create directory /var

8<-

and that's it the graphical diff program never pops up.

when omitting the `--to' everything works just fine.

don't know when it stopped working -- might not have used it for some time  
but it sure used to work. incidentally, in the above example `--to' is  
pointing to the CURRENT revision anyway so it should do exactly the same  
w/ and w/o the `--to' flag...


seen with

This is fossil version 2.3 [2a615bed11] 2017-07-31 17:42:57 UTC under  
macosx 10.12


am I missing something trivial?

thanks
joerg

--
Using Opera's revolutionary email client: http://www.opera.com/mail/
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users