Missing file in EMACS_22_BASE (GIT repo)

Missing file in EMACS_22_BASE (GIT repo)

[dhruva]

Hi,
 I am using the GIT repo. I am not finding the emacs.manifest file
under the 'emacs/nt' folder. The build (MinGW) on XP needs it. I just
copied it from the HEAD and the build went fine. I am able to see this
file in the ViewCVS frontend under the EMACS_22_BASE branch. I am
guessing something might have gone wrong in the GIT repository.

-dky

-- 
Dhruva Krishnamurthy
Contents reflect my personal views only!

Re: Missing file in EMACS_22_BASE (GIT repo)

[Jim Meyering]

dhruva <dhruvakm@gmail.com> wrote:
>  I am using the GIT repo. I am not finding the emacs.manifest file
> under the 'emacs/nt' folder. The build (MinGW) on XP needs it. I just
> copied it from the HEAD and the build went fine. I am able to see this
> file in the ViewCVS frontend under the EMACS_22_BASE branch. I am
> guessing something might have gone wrong in the GIT repository.

If you don't have it, then perhaps you removed it or interrupted the clone.
It's certainly in my copy (obtained via git clone git://git.sv.gnu.org/emacs):

  $ git log nt/emacs.manifest
  commit f48fe6668ba687f79be000712e2c0df709a7a7f0
  Author: Jason Rumney <jasonr@gnu.org>
  Date:   Fri Jun 15 00:07:45 2007 +0000

      Manifest for Windows XP and later. Links Emacs to comctl32.dll version 6 for
      the new style scrollbars.

Re: Missing file in EMACS_22_BASE (GIT repo)

[dhruva]

Hi,

On 10/4/07, Jim Meyering <jim@meyering.net> wrote:
> dhruva <dhruvakm@gmail.com> wrote:
> >  I am using the GIT repo. I am not finding the emacs.manifest file
> > under the 'emacs/nt' folder. The build (MinGW) on XP needs it. I just
> > copied it from the HEAD and the build went fine. I am able to see this
> > file in the ViewCVS frontend under the EMACS_22_BASE branch. I am
> > guessing something might have gone wrong in the GIT repository.
>
> If you don't have it, then perhaps you removed it or interrupted the clone.
> It's certainly in my copy (obtained via git clone git://git.sv.gnu.org/emacs):

Strange.. I do not see it in the gitweb either:
http://git.sv.gnu.org/gitweb/?p=emacs.git;a=tree;f=nt;h=844d19ca974ab7b2b505ef21f77b138862c7e4ad;hb=EMACS_22_BASE


-dky

-- 
Dhruva Krishnamurthy
Contents reflect my personal views only!

Re: Missing file in EMACS_22_BASE (GIT repo)

[Jim Meyering]

dhruva <dhruvakm@gmail.com> wrote:
> On 10/4/07, Jim Meyering <jim@meyering.net> wrote:
>> dhruva <dhruvakm@gmail.com> wrote:
>> >  I am using the GIT repo. I am not finding the emacs.manifest file
>> > under the 'emacs/nt' folder. The build (MinGW) on XP needs it. I just
>> > copied it from the HEAD and the build went fine. I am able to see this
>> > file in the ViewCVS frontend under the EMACS_22_BASE branch. I am
>> > guessing something might have gone wrong in the GIT repository.
>>
>> If you don't have it, then perhaps you removed it or interrupted the clone.
>> It's certainly in my copy (obtained via git clone git://git.sv.gnu.org/emacs):
>
> Strange.. I do not see it in the gitweb either:
> http://git.sv.gnu.org/gitweb/?p=emacs.git;a=tree;f=nt;h=844d19ca974ab7b2b505ef21f77b138862c7e4ad;hb=EMACS_22_BASE

I see what you mean, now.
I missed the part about "on a branch".

If you can confirm that that's the only file, it probably makes
sense simply to add it manually on that branch.

Re: Missing file in EMACS_22_BASE (GIT repo)

[dhruva]

Hi,

On 10/4/07, Jim Meyering <jim@meyering.net> wrote:
> I see what you mean, now.
> I missed the part about "on a branch".
>
> If you can confirm that that's the only file, it probably makes
> sense simply to add it manually on that branch.

I will checkout from CVS and do a directory diff and send my findings
tomorrow, quite late now, have ot go home :-)

-dky

-- 
Dhruva Krishnamurthy
Contents reflect my personal views only!

Re: Missing file in EMACS_22_BASE (GIT repo)

[dhruva]

Hi,

On 10/4/07, dhruva <dhruvakm@gmail.com> wrote:
> Hi,
>
> On 10/4/07, Jim Meyering <jim@meyering.net> wrote:
> > I see what you mean, now.
> > I missed the part about "on a branch".
> >
> > If you can confirm that that's the only file, it probably makes
> > sense simply to add it manually on that branch.


I confirm that 'nt/emacs.manifest' is the only missing file. However,
I notice a difference in a few files. In the following unified diffs
(does emacs devel prefer unified/contextual diffs?), the first is the
CVS repo and the second is the GIT repo (CVS vs GIT comparison).

--- ./emacs/lisp/ChangeLog      Thu Oct  4 22:10:33 2007
+++ /e/users/dky/stub/hacks/./emacs/lisp/ChangeLog      Thu Oct  4 17:43:06 2007
@@ -1,7 +1,3 @@
-2007-10-04  Juanma Barranquero  <lekktu@gmail.com>
-
-       * image-dired.el (image-dired-image-at-point-p): Fix typo in docstring.
-
 2007-10-04  MichaAkBl Cadilhac  <michael@cadilhac.name>

        * progmodes/gud.el (gud-gud-gdb-command-name): Fix typo in docstring.
--- ./emacs/lisp/image-dired.el Thu Oct  4 22:05:17 2007
+++ /e/users/dky/stub/hacks/./emacs/lisp/image-dired.el Thu Oct  4 17:43:12 2007
@@ -1851,7 +1851,7 @@
   (image-dired-display-image (dired-get-filename) arg))

 (defun image-dired-image-at-point-p ()
-  "Return true if there is an image-dired thumbnail at point."
+  "Return true if there is a image-dired thumbnail at point."
   (get-text-property (point) 'image-dired-thumbnail))

 (defun image-dired-rotate-thumbnail (degrees)
--- ./emacs/src/ChangeLog       Fri Oct  5 06:01:33 2007
+++ /e/users/dky/stub/hacks/./emacs/src/ChangeLog       Thu Oct  4 17:43:24 2007
@@ -12,10 +12,6 @@
        * emacs.c (standard_args): Change priority of "--no-splash"
        from 40 to 3.  Add "--no-desktop" with the same priority.

-2007-09-29  Jason Rumney  <jasonr@gnu.org>
-
-       * w32.c (w32_get_resource): Always close registry keys.
-


-dky

-- 
Dhruva Krishnamurthy
Contents reflect my personal views only!

Re: Missing file in EMACS_22_BASE (GIT repo)

[Jason Rumney]

dhruva wrote:
> -2007-10-04  Juanma Barranquero  <lekktu@gmail.com>
> -
> -       * image-dired.el (image-dired-image-at-point-p): Fix typo in docstring.
> -
> -2007-09-29  Jason Rumney  <jasonr@gnu.org>
> -
> -       * w32.c (w32_get_resource): Always close registry keys.
> -
>   

These are both recent changes, so the explanation is that they haven't
yet been synced. The second had failed earlier, and I only noticed
yesterday.

Re: Missing file in EMACS_22_BASE (GIT repo)

[Andreas Schwab]

Jim Meyering <jim@meyering.net> writes:

> dhruva <dhruvakm@gmail.com> wrote:
>> On 10/4/07, Jim Meyering <jim@meyering.net> wrote:
>>> dhruva <dhruvakm@gmail.com> wrote:
>>> >  I am using the GIT repo. I am not finding the emacs.manifest file
>>> > under the 'emacs/nt' folder. The build (MinGW) on XP needs it. I just
>>> > copied it from the HEAD and the build went fine. I am able to see this
>>> > file in the ViewCVS frontend under the EMACS_22_BASE branch. I am
>>> > guessing something might have gone wrong in the GIT repository.
>>>
>>> If you don't have it, then perhaps you removed it or interrupted the clone.
>>> It's certainly in my copy (obtained via git clone git://git.sv.gnu.org/emacs):
>>
>> Strange.. I do not see it in the gitweb either:
>> http://git.sv.gnu.org/gitweb/?p=emacs.git;a=tree;f=nt;h=844d19ca974ab7b2b505ef21f77b138862c7e4ad;hb=EMACS_22_BASE
>
> I see what you mean, now.
> I missed the part about "on a branch".

That looks like a bug in cvsps, it is already missing from its output.
There are also some tags missing.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Missing file in EMACS_22_BASE (GIT repo)

[Andreas Schwab]

Andreas Schwab <schwab@suse.de> writes:

> Jim Meyering <jim@meyering.net> writes:
>
>> dhruva <dhruvakm@gmail.com> wrote:
>>> On 10/4/07, Jim Meyering <jim@meyering.net> wrote:
>>>> dhruva <dhruvakm@gmail.com> wrote:
>>>> >  I am using the GIT repo. I am not finding the emacs.manifest file
>>>> > under the 'emacs/nt' folder. The build (MinGW) on XP needs it. I just
>>>> > copied it from the HEAD and the build went fine. I am able to see this
>>>> > file in the ViewCVS frontend under the EMACS_22_BASE branch. I am
>>>> > guessing something might have gone wrong in the GIT repository.
>>>>
>>>> If you don't have it, then perhaps you removed it or interrupted the clone.
>>>> It's certainly in my copy (obtained via git clone git://git.sv.gnu.org/emacs):
>>>
>>> Strange.. I do not see it in the gitweb either:
>>> http://git.sv.gnu.org/gitweb/?p=emacs.git;a=tree;f=nt;h=844d19ca974ab7b2b505ef21f77b138862c7e4ad;hb=EMACS_22_BASE
>>
>> I see what you mean, now.
>> I missed the part about "on a branch".
>
> That looks like a bug in cvsps, it is already missing from its output.

Looking closer the situation is more complicated.  The file was added to
the trunk on 2007-06-15, but the EMACS_22_BASE branch was cut on
2007-04-27.  There is no indication in the RCS file how the file was
added to the EMACS_22_BASE branch (unlike the emacs-unicode-2,
multi-tty, unicode-xft and lexbind branches, which have explicit
revisions for the merges).  It looks like the history of the file has
been modified after the fact, and neither cvsps nor git-cvsimport can be
blamed here.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Missing file in EMACS_22_BASE (GIT repo)

[Jason Rumney]

Andreas Schwab wrote:
> Looking closer the situation is more complicated.  The file was added to
> the trunk on 2007-06-15, but the EMACS_22_BASE branch was cut on
> 2007-04-27.  There is no indication in the RCS file how the file was
> added to the EMACS_22_BASE branch (unlike the emacs-unicode-2,
> multi-tty, unicode-xft and lexbind branches, which have explicit
> revisions for the merges).  It looks like the history of the file has
> been modified after the fact, and neither cvsps nor git-cvsimport can be
> blamed here.
>   


It was added to the branches by tagging, which is the proper way to do
such things in CVS. It seems it was then deleted from the other branches
and re-added, presumably for the benefit of other revision control
systems with which CVS is synched that have different ideas about what a
branch is.

Re: Missing file in EMACS_22_BASE (GIT repo)

[Andreas Schwab]

Jason Rumney <jasonr@gnu.org> writes:

> Andreas Schwab wrote:
>> Looking closer the situation is more complicated.  The file was added to
>> the trunk on 2007-06-15, but the EMACS_22_BASE branch was cut on
>> 2007-04-27.  There is no indication in the RCS file how the file was
>> added to the EMACS_22_BASE branch (unlike the emacs-unicode-2,
>> multi-tty, unicode-xft and lexbind branches, which have explicit
>> revisions for the merges).  It looks like the history of the file has
>> been modified after the fact, and neither cvsps nor git-cvsimport can be
>> blamed here.
>>   
>
> It was added to the branches by tagging, which is the proper way to do
> such things in CVS.

The file didn't exist yet when the EMACS_22_BASE branch was created.

> It seems it was then deleted from the other branches and re-added,

This is done automatically by CVS when merging a new file from a
different branch.  This makes sure the file is marked dead before the
merge.  This revision is missing for the EMACS_22_BASE branch.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Missing file in EMACS_22_BASE (GIT repo)

[Jason Rumney]

Andreas Schwab wrote:
> This is done automatically by CVS when merging a new file from a
> different branch.  This makes sure the file is marked dead before the
> merge.  This revision is missing for the EMACS_22_BASE bran
I don't see why a file should be marked dead before it is even born.
Commands such as "cvs update -rEMACS_22_BRANCH -D2007-06-14" and "cvs
update -rEMACS_22_1" still work as they should without cluttering the
repository with redundant versions of files.

Re: Missing file in EMACS_22_BASE (GIT repo)

[Andreas Schwab]

Jason Rumney <jasonr@gnu.org> writes:

> Andreas Schwab wrote:
>> This is done automatically by CVS when merging a new file from a
>> different branch.  This makes sure the file is marked dead before the
>> merge.  This revision is missing for the EMACS_22_BASE bran
> I don't see why a file should be marked dead before it is even born.

It isn't.  The file exists, but only on one branch.  When it is added to
a different branch there needs to be a way to tell that the file didn't
exist before that point on that branch.  That's the way CVS works.  The
current situation cannot have been created without rewriting history.

> Commands such as "cvs update -rEMACS_22_BRANCH -D2007-06-14"
> and "cvs update -rEMACS_22_1" still work as they should without
> cluttering the repository with redundant versions of files.

Now try -rEMACS_22_BRANCHPOINT.  The file magically appears before it
was even created.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Missing file in EMACS_22_BASE (GIT repo)

[Jason Rumney]

Andreas Schwab wrote:
> It isn't.  The file exists, but only on one branch.  When it is added to
> a different branch there needs to be a way to tell that the file didn't
> exist before that point on that branch.  That's the way CVS works.  The
> current situation cannot have been created without rewriting history.
>   
The creation date of that file is what tells that it didn't exist before
then. This is no different on a branch than the trunk. I think you are
thinking of the way git or arch works, not the way CVS works. In CVS,
files branch, not repositories.

> Now try -rEMACS_22_BRANCHPOINT.  The file magically appears before it
> was even created.
>   
The EMACS_22_BRANCHPOINT tag is supposed to be there as an aid to
automated merging. It is not intended to give a snapshot of the
development tree at a specific time. If it is supposed to do the latter,
then it should be removed from that file, since the file did not exist
at that time. The other branchpoint tags such as emacs-unicode-2-base
also present the same problem, regardless of the fact that those
branches contain a redundant deletion and readdition of the file.

Re: Missing file in EMACS_22_BASE (GIT repo)

[Andreas Schwab]

Jason Rumney <jasonr@gnu.org> writes:

> Andreas Schwab wrote:
>> It isn't.  The file exists, but only on one branch.  When it is added to
>> a different branch there needs to be a way to tell that the file didn't
>> exist before that point on that branch.  That's the way CVS works.  The
>> current situation cannot have been created without rewriting history.
>>   
> The creation date of that file is what tells that it didn't exist before
> then.

That is only true for the trunk.  On a branch, there is *always* an
extra commit with the automatically generated commit message "file
... was initially added on branch ..." (optionally followed by the date
of the merge when the file was first added on the trunk).

> This is no different on a branch than the trunk. I think you are
> thinking of the way git or arch works, not the way CVS works. In CVS,
> files branch, not repositories.

The current situation cannot be created without directly manipulating
the branch tag.

>> Now try -rEMACS_22_BRANCHPOINT.  The file magically appears before it
>> was even created.
>>   
> The EMACS_22_BRANCHPOINT tag is supposed to be there as an aid to
> automated merging.

No.  The tag is supposed to represent the branch point exactly.

> The other branchpoint tags such as emacs-unicode-2-base also present
> the same problem,

The emacs-unicode-2-base tag is indeed a moving tag.

> regardless of the fact that those branches contain a redundant
> deletion and readdition of the file.

They are not redundant, those revisions are a direct result by the use
of proper CVS operation.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Missing file in EMACS_22_BASE (GIT repo)

[Jason Rumney]

Andreas Schwab wrote:
>> The EMACS_22_BRANCHPOINT tag is supposed to be there as an aid to
>> automated merging.
>>     
>
> No.  The tag is supposed to represent the branch point exactly.
>   

Right. It represents the point where each file was branched from the
trunk, as an aid to automated merging. It does not represent a release,
or a snapshot in time. This is no different than the
emacs-unicode-2-base tag, which as I pointed out is also pointing at
revision 1.1 of the file, despite the different mechanism used to get it
onto the branch.

> The emacs-unicode-2-base tag is indeed a moving tag.
>   

It's not moving, it is fixed on version 1.1, the branchpoint for the
emacs-unicode-2 branch. There is then a deleted revision (1.1.2.1) on
the emacs-unicode-2 branch, with the same timestamp as the original
addition of version 1.1, so the only way to get revision 1.1 on that
branch is to use the emacs-unicode-2-base tag. Then 46 hours later, when
Miles did the automated merge, the file comes back into existence.  CVS
does it this way if you later do a "cvs add", because CVS assumes that
for the previous 46 hours, the file should not exist on the branch. But
in the case of the EMACS_22_BASE branch, there is no automatic merge, so
I manually added it at the same time as I added the trunk. Since there
was no delay, there is no need for a deleted revision on the branch.

Re: Missing file in EMACS_22_BASE (GIT repo)

[Andreas Schwab]

Jason Rumney <jasonr@gnu.org> writes:

> Andreas Schwab wrote:
>>> The EMACS_22_BRANCHPOINT tag is supposed to be there as an aid to
>>> automated merging.
>>>     
>>
>> No.  The tag is supposed to represent the branch point exactly.
>>   
>
> Right. It represents the point where each file was branched from the
> trunk, as an aid to automated merging.

It has nothing to do with merging.  It is *the* branch point and
represents the tree directly *before* branching.  The recommended
procedure for branching is to create a tag and _then_ create the branch
based on this tag.

>> The emacs-unicode-2-base tag is indeed a moving tag.
>>   
>
> It's not moving, it is fixed on version 1.1, the branchpoint for the
> emacs-unicode-2 branch.

That revision never existed on that branch, only on the trunk.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."