Re: [fossil-users] [8c22e1bbcd8ec048] Regression: "fossil amend" no longer works unless operating from within an open checkout

2017-05-25 Thread Andy Bradford
Thus said fos...@rkeene.org on Thu, 25 May 2017 23:42:22 -0500:

> Why is the check-out directory considered at all ?

My guess is that it was an expedient fix to prevent a segfault.

I agree  with you that  Fossil shouldn't care  about whether or  not the
repository is open and just choose  a suitable location for the file. If
the  repository is  open, use  the working  check-out directory  for the
temporary file, otherwise use an appropriate TMP location.

Andy
-- 
TAI64 timestamp: 40005927b840


___
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] [8c22e1bbcd8ec048] Regression: "fossil amend" no longer works unless operating from within an open checkout

2017-05-25 Thread fossil
I don't see why Fossil cares if there is an open checkout for editing a 
comment interactively -- create a temporary file in the normal way that 
temporary files are normally created (TMPDIR et. al.) and allow the user 
to edit it, then consume the new contents.


Why is the check-out directory considered at all ?

On Thu, 25 May 2017, Andy Bradford wrote:


Thus said Roy Keene on Thu, 25 May 2017 21:17:15 -0500:


Given that this worked, why was it  broken ? And can the error message
be converted  to something that is  actually coherent ? And  can it be
unbroken ?


Apparently it was broken intentionally because there was a segfault that
happened when using  ``fossil amend -R repo.fossil -e''  outside an open
checkout:

http://marc.info/?t=14897952902=1=2

Is there a reason why one shouldn't be able to use amend outside an open
check-out? One alternative would be to simply disallow interactive edits
outside an open checkout:

http://www.fossil-scm.org/index.html/info/afef5fb5fc49fdd9

However, perhaps this could be improved by detecting whether there is an
open check-out here:

http://www.fossil-scm.org/index.html/artifact?ln=1209-1210=237694d101bab9bc

And  if there  is  no open  check-out use  unixTempFileDir  to locate  a
suitable location and put the file there?

Or maybe  file_relative_name should actually  try to handle a  NULL file
path and simply treat that as  ``current working directory'' and make it
relative to . rather than fail?

Thoughts?

Andy
--
TAI64 timestamp: 40005927a824



___
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] [8c22e1bbcd8ec048] Regression: "fossil amend" no longer works unless operating from within an open checkout

2017-05-25 Thread Andy Bradford
Thus said Roy Keene on Thu, 25 May 2017 21:17:15 -0500:

> Given that this worked, why was it  broken ? And can the error message
> be converted  to something that is  actually coherent ? And  can it be
> unbroken ?

Apparently it was broken intentionally because there was a segfault that
happened when using  ``fossil amend -R repo.fossil -e''  outside an open
checkout:

http://marc.info/?t=14897952902=1=2

Is there a reason why one shouldn't be able to use amend outside an open
check-out? One alternative would be to simply disallow interactive edits
outside an open checkout:

http://www.fossil-scm.org/index.html/info/afef5fb5fc49fdd9

However, perhaps this could be improved by detecting whether there is an
open check-out here:

http://www.fossil-scm.org/index.html/artifact?ln=1209-1210=237694d101bab9bc

And  if there  is  no open  check-out use  unixTempFileDir  to locate  a
suitable location and put the file there?

Or maybe  file_relative_name should actually  try to handle a  NULL file
path and simply treat that as  ``current working directory'' and make it
relative to . rather than fail?

Thoughts?

Andy
-- 
TAI64 timestamp: 40005927a824


___
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] [8c22e1bbcd8ec048] Regression: "fossil amend" no longer works unless operating from within an open checkout

2017-05-25 Thread Andy Bradford
Thus said Roy Keene on Thu, 25 May 2017 21:26:50 -0500:

> The fossil checkin I linked  to ([8c22e1bbcd8ec048]) has the change in
> src/info.c that enforces this for all cases.

Yeah, I think  you already said that  and somehow I missed  it. I wonder
why this was introduced...

Andy
-- 
TAI64 timestamp: 4000592794b9


___
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] [8c22e1bbcd8ec048] Regression: "fossil amend" no longer works unless operating from within an open checkout

2017-05-25 Thread Andy Bradford
Thus said Roy Keene on Thu, 25 May 2017 21:17:15 -0500:

> Attempts to add tags to a repository using "-R" now returns an engrish
> error message:

The engrish error message has been addressed, however, it does look like
somewhere the behavior has changed. I gave  it a valid UUID and it still
gives me the error.

Andy
-- 
TAI64 timestamp: 400059279380


___
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] [8c22e1bbcd8ec048] Regression: "fossil amend" no longer works unless operating from within an open checkout

2017-05-25 Thread Roy Keene

Andy,

	Sorry -- the actual command had the UUID in there, it got stripped 
out during sanitization:

$ fossil amend -R ~autobuild/aurae.fossil 
0fb28f59263f1a0c547c9abebeb69d2da67d3f17 --tag build-pass --tag tests-fail
the "amend" command only work from within an open check-out

The fossil checkin I linked to ([8c22e1bbcd8ec048]) has the change in 
src/info.c that enforces this for all cases.


Thanks,
Roy Keene


On Thu, 25 May 2017, Andy Bradford wrote:


Thus said Roy Keene on Thu, 25 May 2017 21:17:15 -0500:


$ fossil amend -R ~autobuild/aurae.fossil --tag build-pass --tag 
tests-fail
the "amend" command only work from within an open check-out


What particular checkin are you trying to amend here?

Seems like the error really should be a usage statement.

Try:

fossil amend -R ~autobuild/aurae.fossil UUID --tag build-pass --tag tests-fail

Where UUID is the checkin that you want to amend.

Andy
--
TAI64 timestamp: 4000592791c4



___
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] [8c22e1bbcd8ec048] Regression: "fossil amend" no longer works unless operating from within an open checkout

2017-05-25 Thread Andy Bradford
Thus said Roy Keene on Thu, 25 May 2017 21:17:15 -0500:

>   $ fossil amend -R ~autobuild/aurae.fossil --tag build-pass --tag 
> tests-fail
>   the "amend" command only work from within an open check-out

What particular checkin are you trying to amend here?

Seems like the error really should be a usage statement.

Try:

fossil amend -R ~autobuild/aurae.fossil UUID --tag build-pass --tag tests-fail

Where UUID is the checkin that you want to amend.

Andy
-- 
TAI64 timestamp: 4000592791c4


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


[fossil-users] [8c22e1bbcd8ec048] Regression: "fossil amend" no longer works unless operating from within an open checkout

2017-05-25 Thread Roy Keene

All,

Per the change made here:
http://fossil-scm.org/index.html/info/8c22e1bbcd8ec048

Attempts to add tags to a repository using "-R" now returns an engrish 
error message:


$ fossil amend -R ~autobuild/aurae.fossil --tag build-pass --tag 
tests-fail
the "amend" command only work from within an open check-out

Given that this worked, why was it broken ?   And can the error message be 
converted to something that is actually coherent ?  And can it be unbroken ?


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