headers
David Brown | 3 Feb 2005 01:18

Cannot cross branch merge.

I cannot figure out how to clone a change on a different branch.

   aeclone -p project.1.2 -br 1.1 39

When I run 'aed -om', the file pulled out of the history for $original does
not appear to be a common ancestor, but instead appears to be identical to
$input.  This causes the merge utility to undo all of the changes that were
made.

Do we know that $original is doing the right thing in this cross-branch
clone.

As long as there isn't any real merging, moving the ,B files onto the main
file puts the changes in place.  But, if there really is any merging, the
right common ancestor doesn't appear to be available.

Dave
Permalink | Reply | 1 comment |
headers
Brendan Simon | 3 Feb 2005 02:44

Re: Cannot cross branch merge.

I came accross this problem or a similar one.
Are you trying to clone a branch or clone a changeset ???

I use aepatch to merge changesets on one branch to a new changeset on 
another branch.

I use aeclone to copy a branch to a new branch (but there are not 
ancestor files, except aegis config files)
eg. move from version 1.2.3 to 1.2.4.
$ aeclone -p foo.1.2.4 -br 1.2 -c 3

Brendan Simon.

David Brown wrote:

>I cannot figure out how to clone a change on a different branch.
>
>   aeclone -p project.1.2 -br 1.1 39
>
>When I run 'aed -om', the file pulled out of the history for $original does
>not appear to be a common ancestor, but instead appears to be identical to
>$input.  This causes the merge utility to undo all of the changes that were
>made.
>
>Do we know that $original is doing the right thing in this cross-branch
>clone.
>
>As long as there isn't any real merging, moving the ,B files onto the main
>file puts the changes in place.  But, if there really is any merging, the
>right common ancestor doesn't appear to be available.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Brendan Simon | 4 Feb 2005 04:46

Change file to have build attribute

I have a number of files in a project that I want to change to have the 
build attribute.
Is aefa the appropriate tool to do the job, or do I have to do some kind 
fancy aerm followed by aenf -build ???

I have aegis.4.16 with C013 patched in.  It looks like aefa was 
introduced in 4.17 so I guess I should upgrade to 4.20, assuming there 
are no reports of major problems with 4.20.

Ta,
Brendan Simon.
Permalink | Reply | 0 comments |
headers
Ralf Fassel | 8 Feb 2005 18:15
Picon

Re: Aegis 4.20 - a project change supervisor

* Peter Miller
| I am pleased to announce that Aegis 4 is now available.

aegis 4.20, compilation overview:

- The BUILDING file should mention CXX and CXXFLAGS at the same place
  as CC and CFLAGS

Summary:
- compiles and tests ok on
  - linux 2.4/Redhat WS3, gcc 3.2.3
  - linux 2.6/Suse 9.1, gcc 3.3.3
  - IRIX 6.5.22m, gcc 3.4.3, set DIFFsimple in environment before
    running `configure'

- compilation failure
  - on HP-10.20, ICE in gcc 3.4.1 and gcc 3.4.3
  - on IRIX 6.5.22m, gcc 3.4.1 due to an header file error in gcc

Details:  
======================================================================================
* Linux ozelot 2.6.5-7.111.30-default #1 Fri Jan 14 12:58:46 UTC 2005 i686 i686 i386 GNU/Linux
no libcurl found
rpm -qa | grep -i curl => curl-7.11.0-39
Passed All Tests
======================================================================================

======================================================================================
* Linux calypso 2.4.21-27.0.2.EL #1 Wed Jan 12 23:35:46 EST 2005 i686 athlon i386 GNU/Linux
libcurl installed
(Continue reading)

Permalink | Reply | 0 comments |
headers
Simon Clift | 9 Feb 2005 17:55
Picon
Picon
Favicon

Aegis 4.20/Cook 2.25 : strange problem with non-existent "state" file

Hello fellow Aegis users,

I've just updated my Cook and Aegis installations to 2.25
and 4.20 respectively.  I've run into a problem with aeb and
cook, and would like to ask for some help, please, in
finding a solution.

After aeb, my Howto.list reports:

/* Howto.list, /home/ssclift/Projects/sjopt.1.2.C013, Wed Feb 09 11:43
2005 */
cook: date +"%Y-%m-%d %H:%M:%S"
cook: aegis -change_directory -baseline
aegis: /home/ssclift/Projects/sjopt.1.2.C013/state: Permission denied
[user "adm", group "sys"]
cook: aegis: exit status 1
cook: Howto.cook: 20: statement failed
cook: Howto.cook: found 1 fatal error

and aegis.log just reports that cook failed.

The statement that fails is the line

  "aegis -change_directory -baseline"

and other aegis command fail similarly if I comment that
line out.  The ".../sjopt.1.2.C013/state" file in question
does not exist.  Creating it does not help, adding a+w
permissions on the project change directory doesn't help.
(Continue reading)

Permalink | Reply | 3 comments |
headers
Simon Clift | 9 Feb 2005 19:53
Picon
Picon
Favicon

Re: Aegis 4.20/Cook 2.25 : strange problem with non-existent "state" file

On Wed, 2005-02-09 at 11:55 -0500, Simon Clift wrote:
> After aeb, my Howto.list reports:
...
> aegis: /home/ssclift/Projects/sjopt.1.2.C013/state: Permission denied
> [user "adm", group "sys"]
> cook: aegis: exit status 1

Hmmm... my VMS colleagues always complain about the
magic and cryptic incantations required to make 
Unix go...  I found a spell that works on my problem.

I set the AEGIS_PATH environment variable and the 
problem went away.

There's a slight whiff of burnt wolfsbane in the air, 
though...  :-)

Best regards
-- Simon

--

-- 
Simon Clift                        ssclift at math dot uwaterloo dot ca
Ph.D. Student, Computational Finance
Scientific Computing Group, School of Computer Science
University of Waterloo
Waterloo, Ontario, Canada   N2L 3G1
Permalink | Reply | 2 comments |
headers
Jerry Pendergraft | 10 Feb 2005 04:14

Re: Aegis 4.20/Cook 2.25 : strange problem with non-existent "state" file

Something seems verry fishy with this. There is no way that aegis should
be looking for the state file in what looks like a change directory.
Did you install aegis with setuid root ownership and permissions?

-- 
Jerry Pendergraft                        jerry.pendergraft <at> endocardial.com
Endocardial Solutions                    voice: 651-523-6935
1350 Energy Lane, Suite 110                fax: 651-644-7897
St Paul, MN 55108-5254                  mobile: 651-491-0163

On Wed, 9 Feb 2005, Simon Clift wrote:

> On Wed, 2005-02-09 at 11:55 -0500, Simon Clift wrote:
> > After aeb, my Howto.list reports:
> ...
> > aegis: /home/ssclift/Projects/sjopt.1.2.C013/state: Permission denied
> > [user "adm", group "sys"]
> > cook: aegis: exit status 1
> 
> Hmmm... my VMS colleagues always complain about the
> magic and cryptic incantations required to make 
> Unix go...  I found a spell that works on my problem.
> 
> I set the AEGIS_PATH environment variable and the 
> problem went away.
> 
> There's a slight whiff of burnt wolfsbane in the air, 
> though...  :-)
> 
> Best regards
(Continue reading)

Permalink | Reply | 1 comment |
headers
Simon Clift | 10 Feb 2005 16:56
Picon
Picon
Favicon

Re: Aegis 4.20/Cook 2.25 : strange problem with non-existent "state" file

On Wed, 2005-02-09 at 21:14 -0600, Jerry Pendergraft wrote:
> Something seems verry fishy with this. There is no way that aegis should
> be looking for the state file in what looks like a change directory.
> Did you install aegis with setuid root ownership and permissions?

Yup, like our local river after the smelt run... :-)

Aegis is indeed installed setuid root.

ssclift[999] ls -l /usr/bin/aegis
-rwsr-xr-x  1 root bin 1293472 Feb  8 21:55 /usr/bin/aegis

All of the other things I've done with the new version of  Aegis
have showed no problems.  I'm getting close to a big change test
and check-in... I'll see how it goes.

Best regards
-- Simon

--

-- 
Simon Clift                        ssclift at math dot uwaterloo dot ca
Ph.D. Student, Computational Finance
Scientific Computing Group, School of Computer Science
University of Waterloo
Waterloo, Ontario, Canada   N2L 3G1
Permalink | Reply | 0 comments |
headers
Pieter Nagel | 11 Feb 2005 10:15
Picon

Cycle in file move history, can't aedist baseline

Aegis seems to misinterpret an aemv followed by an aerm of the new name as
some sort of cycle; and no I can't aedist my project baseline.

In short: file A had been renamed to B, and then, in a later change C, B had
been removed. ael pf reports that A had been renamed to B, and that B had
been renamed to A, and that both had been deleted in change C. aedist tries
to find file A in the baseline.

The failing aedist command:

aedist -send -baseline -compress -output ~/dante.ae -project dante.3.51
aedist: open "/net/pixar/aegis/dante/branch.3/baseline/OldStyleTests/learn/WithdrawalTests_checkInconsistencyWithPortfolio1"

The output of "ael pf -p dante.3", just for the offending file:

source 	remove 	 1.3	OldStyleTests/WithdrawalTests_checkInconsistencyWithPortfolio-1.html.learn
			Deleted by change 46.
			Moved to OldStyleTests/learn/WithdrawalTests_checkInconsistencyWithPortfolio1
source 	remove 	 1.1	OldStyleTests/learn/WithdrawalTests_checkInconsistencyWithPortfolio1
			Deleted by change 46.
			Moved to OldStyleTests/WithdrawalTests_checkInconsistencyWithPortfolio-1.html.learn

This is wrong. What actually happened is that "*/learn/*" was aemv'ed to
"*/*.learn", and some of the latter were later aerm'ed in change 46.

--
  , _
   /_)           /| /
  /  i e t e r  / |/ a g e l
(Continue reading)

Permalink | Reply | 3 comments |
headers
Jerry Pendergraft | 11 Feb 2005 16:33

Re: Cycle in file move history, can't aedist baseline

I believe you can work around the problem for now by touching the missing
file[s] so aedist can finish. Then it would be very interesting to see
what aedist -receive on a remote or different project would produce.

-- 
Jerry Pendergraft                        jerry.pendergraft <at> endocardial.com
Endocardial Solutions                    voice: 651-523-6935
1350 Energy Lane, Suite 110                fax: 651-644-7897
St Paul, MN 55108-5254                  mobile: 651-491-0163

On Fri, 11 Feb 2005, Pieter Nagel wrote:

> Aegis seems to misinterpret an aemv followed by an aerm of the new name as
> some sort of cycle; and no I can't aedist my project baseline.
> 
> In short: file A had been renamed to B, and then, in a later change C, B had
> been removed. ael pf reports that A had been renamed to B, and that B had
> been renamed to A, and that both had been deleted in change C. aedist tries
> to find file A in the baseline.
> 
> 
> The failing aedist command:
> 
> aedist -send -baseline -compress -output ~/dante.ae -project dante.3.51
> aedist: open "/net/pixar/aegis/dante/branch.3/baseline/OldStyleTests/learn/WithdrawalTests_checkInconsistencyWithPortfolio1"
> 
> 
> The output of "ael pf -p dante.3", just for the offending file:
> 
> source 	remove 	 1.3	OldStyleTests/WithdrawalTests_checkInconsistencyWithPortfolio-1.html.learn
(Continue reading)

Permalink | Reply | 2 comments |

Gmane