Re: Any recommendations for off-line CVS/egcs work?

Re: Any recommendations for off-line CVS/egcs work?

[Mike Stump]

> From: craig@jcb-sc.com
> Date: 4 Feb 1999 18:38:52 -0000
> To: egcs@egcs.cygnus.com

> If anyone has pointers to information on, or boilerplate/sample
> scripts, Emacs lisp codes, easing doing egcs work (via CVS) largely
> off-line, please drop me a note.

With just a little work, cvs allows for this type of operation quite
well.  I did it for years from a dialup with cvs (going back to a time
before remote cvs even).  I'll post here, since there may still be
folks in Europe or other such places that prefer an offline mode as
well.

> every CVS operation

?  I didn't use all that many, I still don't.  I wish you had listed
them so I could cover them.  At the end is mkdiff, which will given
either a list of *~ files (or ~1~ files) or directories, will produce
a diff.  The next bit of magic is .emacs fodder to created numbered
backups:

(defun i-want-version-control () (setq version-control t))
(setq keep-old-versions 5)
(setq keep-new-versions 5)
(setq kept-old-versions 50)
(setq kept-new-versions 50)

and the a script, called vmacs:

#!/bin/sh
#DISPLAY=$DISPLAY
#export DISPLAY
unset DISPLAY
exec emacs -f i-want-version-control ${1+"$@"}

that sets the default to be numbered backups in edit sessions.  You
have to handle changelogs specially this way (I just put the into the
top of my patches file, and the mkdiff . >>patch-2.diffs).

I start with an up-to-date tree once a week (or once a day, or once
per unit of work), and just do them up one by one, saving them off.
After saving them off, I do a mkorig (see below) to restore the tree
to a virgin state to start the next batch of work.  Be sure to remove
*~ after you do a mkorig.  It is also useful in finding _what_ I have
changed so that I know what to check in.  I run a cvs annotate on
*.[chy] and squirrel the output away for reference, updating this
about once a month.

Also, if you patch things, use patch -Vnumbered exclusively.

I have a 10mbps connection now at home, complete with a full mirror of
the egcs cvs repository, and I still use mkdiff/mkorig!

Personally, if you have 100 hours, you should script bringing up and
tearing down your connection and just connect and disconnect and do
live checkins and cvs updates.  You can do up a weeks worth of work,
then cvs update and cvs ci.  This should work well.  I prefer that
over the diffing and patching.  The diffing and patching goes back to
a time before remote cvs, with remote cvs, mkdiff is useful for
reviewing work, and creating changelog entries.


#!/bin/bash
# mkdiff [directory...]
# The below is for function name inclusion in GNU diff.
P=-p
export P

if [ $# = 0 ]; then
	set .
fi
for i
do
	echo "Doing diffs in $i:"
#	find "$i" -name "*.~0~" -print | 
#	while read n
#	do
#		diff -c $P "$n" "${n%.~[01]~}"
#	done
	find "$i" -name ".*.~1~" -print -o -name "*.~1~" -print | 
	while read n
	do
		if [ ! -r "${n%.~1~}.~0~" ]; then
			diff -c $P "$n" "${n%.~[01]~}"
		fi
	done
	echo "--------------"
done


#!/bin/bash
# mkorig
# Move original editor backup to file.
if [ $# = 0 ]; then
    find . -name .\*.~1~ -print -o -name \*.~1~ -print | while read i
    do
	echo mv $i ${i%.~1~}
    done
    # find . -name \*~ | xargs rm -f
else
    for i in ${1+"$@"}; do
	case "$i" in
	*.~1~)
	    echo mv $i ${i%.~1~};;
	esac
    done
    # find . -name \*~ | xargs rm -f
fi


Hope this helps...  If you (or anyone else) have any other concerns or
questions, I'd be happy to answer them.

Re: Any recommendations for off-line CVS/egcs work?

[Dave Love]

>>>>> "Mike" == Mike Stump <mrs@wrs.com> writes:

 Mike> (defun i-want-version-control () (setq version-control t))
 Mike> (setq keep-old-versions 5)
 Mike> (setq keep-new-versions 5)

What do the above do?

 Mike> (setq kept-old-versions 50)
 Mike> (setq kept-new-versions 50)

 Mike> and the a script, called vmacs:

To do that sort of thing only for egcs files, not globally, I'd put it
on `find-file-hooks' to set local variables, presumably depending on
the directory.  Here's an example in another context:

(add-hook 'change-log-mode-hook
	  (lambda ()
	    (if (string-match "\\(/emacs-\\|/egcs\\b\\)" (buffer-file-name))
		(set (make-local-variable 'add-log-mailing-address)
		     "fx@gnu.org"))))

Re: Any recommendations for off-line CVS/egcs work?

[craig]

[Another old message of mine that was rejected.  The *good news* is
that the SMTP server that sends out my email, 199.172.62.20, is no
longer listed at ORBS, so egcs.cygnus.com is now accepting my emails,
some of which are old.  But this was old enough to have been deleted
before that happened.]


I'm a little late, but wanted to thank you for responding with such
a thorough, and apparently helpful, message!  But I am digesting it
a bit slowly, trying to make sure I really understand what is going on.

>> every CVS operation
>
>?  I didn't use all that many, I still don't.  I wish you had listed
>them so I could cover them.

I think I was thinking more in terms of many instances of just the
main checkout/update/commit operations.

>At the end is mkdiff, which will given
>either a list of *~ files (or ~1~ files) or directories, will produce
>a diff.

mkdiff and mkorig look familiar -- I have my own MAKE-P and REVERT-TO-1
scripts, which are (probably) more primitive and (certainly) CVS-unaware
versions of yours.

>The next bit of magic is .emacs fodder to created numbered
>backups:

Great, I've already got stuff like that in my .emacs.

>You
>have to handle changelogs specially this way (I just put the into the
>top of my patches file, and the mkdiff . >>patch-2.diffs).

I've found that ChangeLog behavior sufficiently annoying that I've
gotten to where I habitually switch into fundamental mode just to
start saving a numbered backup, and then switch back.  I'll probably
try and find a way to automate the inclusion of the pertinent ChangeLog
entry as a non-diff-like header in a patch, at some point.

>Personally, if you have 100 hours, you should script bringing up and
>tearing down your connection and just connect and disconnect and do
>live checkins and cvs updates.  You can do up a weeks worth of work,
>then cvs update and cvs ci.  This should work well.  I prefer that
>over the diffing and patching.  The diffing and patching goes back to
>a time before remote cvs, with remote cvs, mkdiff is useful for
>reviewing work, and creating changelog entries.

100 hours per month does seem like plenty of time.  Though I'd like to
have a neatly automated process, regardless, that allows me to easily
decide whether to commit, and/or whether to email, a particular patch,
separately from actually being "on-line" at the time.  That'd be helpful
for the busier times on the 'net.

>Hope this helps...  If you (or anyone else) have any other concerns or
>questions, I'd be happy to answer them.

Thanks lots!

        tq vm, (burley)

- --EAE06059.918670452/europe.std.com--
------- End of forwarded message -------

Re: Any recommendations for off-line CVS/egcs work?

[Zack Weinberg]

>
>I've found that ChangeLog behavior sufficiently annoying that I've
>gotten to where I habitually switch into fundamental mode just to
>start saving a numbered backup, and then switch back.  I'll probably
>try and find a way to automate the inclusion of the pertinent ChangeLog
>entry as a non-diff-like header in a patch, at some point.

I find this bit of elisp quite helpful.  It just inserts the
boilerplate of a changelog entry into the current buffer.   The
(forward-paragraph -1) bit isn't quite right but it comes close.

zw

(defun insert-change-log-entry ()
  "Insert a change log entry before the paragraph at point."
  (interactive)
  (forward-paragraph -1)
  (insert "\n"
	  (format-time-string "%Y-%m-%d %H:%M" (current-time))
	  " " (rfc822-tz) "  "
	  (user-full-name) "  <" user-mail-address ">\n\n\t* \n")
  (forward-char -1) (indented-text-mode))

Re: Any recommendations for off-line CVS/egcs work?

[Paul Derbyshire]

At 03:16 PM 2/11/99 -0000, you wrote:
>[Another old message of mine that was rejected.  The *good news* is
>that the SMTP server that sends out my email, 199.172.62.20, is no
>longer listed at ORBS, so egcs.cygnus.com is now accepting my emails,
>some of which are old.  But this was old enough to have been deleted
>before that happened.]

ORBS?
Why would the egcs list reject someone's mail?

-- 
   .*.  "Clouds are not spheres, mountains are not cones, coastlines are not
-()  <  circles, and bark is not smooth, nor does lightning travel in a
   `*'  straight line."    -------------------------------------------------
        -- B. Mandelbrot  | http://surf.to/pgd.net
_____________________ ____|________     Paul Derbyshire     pderbysh@usa.net
Programmer & Humanist|ICQ: 10423848|

Re: Any recommendations for off-line CVS/egcs work?

[Joe Buck]

> At 03:16 PM 2/11/99 -0000, you wrote:
> >[Another old message of mine that was rejected.  The *good news* is
> >that the SMTP server that sends out my email, 199.172.62.20, is no
> >longer listed at ORBS, so egcs.cygnus.com is now accepting my emails,
> >some of which are old.  But this was old enough to have been deleted
> >before that happened.]
> 
> ORBS?
> Why would the egcs list reject someone's mail?

It's spam protection: if you're from a site that is a significant source
of spam, more and more of the world will simply cut you off (by refusing
to accept your mail).

Re: Any recommendations for off-line CVS/egcs work?

[craig]

>ORBS?

See < http://www.orbs.org >

>Why would the egcs list reject someone's mail?

To not serve spam to everyone on the list.

I'm happy with it.  A little disruption for myself, a little concern
when my ISP's manager said he thought ORBS was too cavalier (my words)
in listing sites and that he'd try to explain this to Cygnus, but
ORBS seems to have undergone a recent change in management and
direction.  In any case, my ISP is no longer listed, hurrah!

        tq vm, (burley)

Re: Any recommendations for off-line CVS/egcs work?

[craig]

[Another old message of mine that was rejected.  The *good news* is
that the SMTP server that sends out my email, 199.172.62.20, is no
longer listed at ORBS, so egcs.cygnus.com is now accepting my emails,
some of which are old.  But this was old enough to have been deleted
before that happened.]


I'm a little late, but wanted to thank you for responding with such
a thorough, and apparently helpful, message!  But I am digesting it
a bit slowly, trying to make sure I really understand what is going on.

>> every CVS operation
>
>?  I didn't use all that many, I still don't.  I wish you had listed
>them so I could cover them.

I think I was thinking more in terms of many instances of just the
main checkout/update/commit operations.

>At the end is mkdiff, which will given
>either a list of *~ files (or ~1~ files) or directories, will produce
>a diff.

mkdiff and mkorig look familiar -- I have my own MAKE-P and REVERT-TO-1
scripts, which are (probably) more primitive and (certainly) CVS-unaware
versions of yours.

>The next bit of magic is .emacs fodder to created numbered
>backups:

Great, I've already got stuff like that in my .emacs.

>You
>have to handle changelogs specially this way (I just put the into the
>top of my patches file, and the mkdiff . >>patch-2.diffs).

I've found that ChangeLog behavior sufficiently annoying that I've
gotten to where I habitually switch into fundamental mode just to
start saving a numbered backup, and then switch back.  I'll probably
try and find a way to automate the inclusion of the pertinent ChangeLog
entry as a non-diff-like header in a patch, at some point.

>Personally, if you have 100 hours, you should script bringing up and
>tearing down your connection and just connect and disconnect and do
>live checkins and cvs updates.  You can do up a weeks worth of work,
>then cvs update and cvs ci.  This should work well.  I prefer that
>over the diffing and patching.  The diffing and patching goes back to
>a time before remote cvs, with remote cvs, mkdiff is useful for
>reviewing work, and creating changelog entries.

100 hours per month does seem like plenty of time.  Though I'd like to
have a neatly automated process, regardless, that allows me to easily
decide whether to commit, and/or whether to email, a particular patch,
separately from actually being "on-line" at the time.  That'd be helpful
for the busier times on the 'net.

>Hope this helps...  If you (or anyone else) have any other concerns or
>questions, I'd be happy to answer them.

Thanks lots!

        tq vm, (burley)

- --EAE06059.918670452/europe.std.com--
------- End of forwarded message -------

Re: Any recommendations for off-line CVS/egcs work?

[Dave Love]

>>>>> "Mike" == Mike Stump <mrs@wrs.com> writes:

 Mike> (defun i-want-version-control () (setq version-control t))
 Mike> (setq keep-old-versions 5)
 Mike> (setq keep-new-versions 5)

What do the above do?

 Mike> (setq kept-old-versions 50)
 Mike> (setq kept-new-versions 50)

 Mike> and the a script, called vmacs:

To do that sort of thing only for egcs files, not globally, I'd put it
on `find-file-hooks' to set local variables, presumably depending on
the directory.  Here's an example in another context:

(add-hook 'change-log-mode-hook
	  (lambda ()
	    (if (string-match "\\(/emacs-\\|/egcs\\b\\)" (buffer-file-name))
		(set (make-local-variable 'add-log-mailing-address)
		     "fx@gnu.org"))))

Re: Any recommendations for off-line CVS/egcs work?

[Paul Derbyshire]

At 03:16 PM 2/11/99 -0000, you wrote:
>[Another old message of mine that was rejected.  The *good news* is
>that the SMTP server that sends out my email, 199.172.62.20, is no
>longer listed at ORBS, so egcs.cygnus.com is now accepting my emails,
>some of which are old.  But this was old enough to have been deleted
>before that happened.]

ORBS?
Why would the egcs list reject someone's mail?

-- 
   .*.  "Clouds are not spheres, mountains are not cones, coastlines are not
-()  <  circles, and bark is not smooth, nor does lightning travel in a
   `*'  straight line."    -------------------------------------------------
        -- B. Mandelbrot  | http://surf.to/pgd.net
_____________________ ____|________     Paul Derbyshire     pderbysh@usa.net
Programmer & Humanist|ICQ: 10423848|

Re: Any recommendations for off-line CVS/egcs work?

[Zack Weinberg]

>
>I've found that ChangeLog behavior sufficiently annoying that I've
>gotten to where I habitually switch into fundamental mode just to
>start saving a numbered backup, and then switch back.  I'll probably
>try and find a way to automate the inclusion of the pertinent ChangeLog
>entry as a non-diff-like header in a patch, at some point.

I find this bit of elisp quite helpful.  It just inserts the
boilerplate of a changelog entry into the current buffer.   The
(forward-paragraph -1) bit isn't quite right but it comes close.

zw

(defun insert-change-log-entry ()
  "Insert a change log entry before the paragraph at point."
  (interactive)
  (forward-paragraph -1)
  (insert "\n"
	  (format-time-string "%Y-%m-%d %H:%M" (current-time))
	  " " (rfc822-tz) "  "
	  (user-full-name) "  <" user-mail-address ">\n\n\t* \n")
  (forward-char -1) (indented-text-mode))

Re: Any recommendations for off-line CVS/egcs work?

[Joe Buck]

> At 03:16 PM 2/11/99 -0000, you wrote:
> >[Another old message of mine that was rejected.  The *good news* is
> >that the SMTP server that sends out my email, 199.172.62.20, is no
> >longer listed at ORBS, so egcs.cygnus.com is now accepting my emails,
> >some of which are old.  But this was old enough to have been deleted
> >before that happened.]
> 
> ORBS?
> Why would the egcs list reject someone's mail?

It's spam protection: if you're from a site that is a significant source
of spam, more and more of the world will simply cut you off (by refusing
to accept your mail).

Re: Any recommendations for off-line CVS/egcs work?

[craig]

>ORBS?

See < http://www.orbs.org >

>Why would the egcs list reject someone's mail?

To not serve spam to everyone on the list.

I'm happy with it.  A little disruption for myself, a little concern
when my ISP's manager said he thought ORBS was too cavalier (my words)
in listing sites and that he'd try to explain this to Cygnus, but
ORBS seems to have undergone a recent change in management and
direction.  In any case, my ISP is no longer listed, hurrah!

        tq vm, (burley)

Re: Any recommendations for off-line CVS/egcs work?

[Mike Stump]

> From: craig@jcb-sc.com
> Date: 4 Feb 1999 18:38:52 -0000
> To: egcs@egcs.cygnus.com

> If anyone has pointers to information on, or boilerplate/sample
> scripts, Emacs lisp codes, easing doing egcs work (via CVS) largely
> off-line, please drop me a note.

With just a little work, cvs allows for this type of operation quite
well.  I did it for years from a dialup with cvs (going back to a time
before remote cvs even).  I'll post here, since there may still be
folks in Europe or other such places that prefer an offline mode as
well.

> every CVS operation

?  I didn't use all that many, I still don't.  I wish you had listed
them so I could cover them.  At the end is mkdiff, which will given
either a list of *~ files (or ~1~ files) or directories, will produce
a diff.  The next bit of magic is .emacs fodder to created numbered
backups:

(defun i-want-version-control () (setq version-control t))
(setq keep-old-versions 5)
(setq keep-new-versions 5)
(setq kept-old-versions 50)
(setq kept-new-versions 50)

and the a script, called vmacs:

#!/bin/sh
#DISPLAY=$DISPLAY
#export DISPLAY
unset DISPLAY
exec emacs -f i-want-version-control ${1+"$@"}

that sets the default to be numbered backups in edit sessions.  You
have to handle changelogs specially this way (I just put the into the
top of my patches file, and the mkdiff . >>patch-2.diffs).

I start with an up-to-date tree once a week (or once a day, or once
per unit of work), and just do them up one by one, saving them off.
After saving them off, I do a mkorig (see below) to restore the tree
to a virgin state to start the next batch of work.  Be sure to remove
*~ after you do a mkorig.  It is also useful in finding _what_ I have
changed so that I know what to check in.  I run a cvs annotate on
*.[chy] and squirrel the output away for reference, updating this
about once a month.

Also, if you patch things, use patch -Vnumbered exclusively.

I have a 10mbps connection now at home, complete with a full mirror of
the egcs cvs repository, and I still use mkdiff/mkorig!

Personally, if you have 100 hours, you should script bringing up and
tearing down your connection and just connect and disconnect and do
live checkins and cvs updates.  You can do up a weeks worth of work,
then cvs update and cvs ci.  This should work well.  I prefer that
over the diffing and patching.  The diffing and patching goes back to
a time before remote cvs, with remote cvs, mkdiff is useful for
reviewing work, and creating changelog entries.


#!/bin/bash
# mkdiff [directory...]
# The below is for function name inclusion in GNU diff.
P=-p
export P

if [ $# = 0 ]; then
	set .
fi
for i
do
	echo "Doing diffs in $i:"
#	find "$i" -name "*.~0~" -print | 
#	while read n
#	do
#		diff -c $P "$n" "${n%.~[01]~}"
#	done
	find "$i" -name ".*.~1~" -print -o -name "*.~1~" -print | 
	while read n
	do
		if [ ! -r "${n%.~1~}.~0~" ]; then
			diff -c $P "$n" "${n%.~[01]~}"
		fi
	done
	echo "--------------"
done


#!/bin/bash
# mkorig
# Move original editor backup to file.
if [ $# = 0 ]; then
    find . -name .\*.~1~ -print -o -name \*.~1~ -print | while read i
    do
	echo mv $i ${i%.~1~}
    done
    # find . -name \*~ | xargs rm -f
else
    for i in ${1+"$@"}; do
	case "$i" in
	*.~1~)
	    echo mv $i ${i%.~1~};;
	esac
    done
    # find . -name \*~ | xargs rm -f
fi


Hope this helps...  If you (or anyone else) have any other concerns or
questions, I'd be happy to answer them.

Any recommendations for off-line CVS/egcs work?

[craig]

If anyone has pointers to information on, or boilerplate/sample scripts,
Emacs lisp codes, easing doing egcs work (via CVS) largely off-line,
please drop me a note.

Specifically, I have dial-up access at an hourly charge.  (Actually,
I get the first 100 hours per month for a flat rate, so the cost isn't
so much the problem as the inconvenience of connecting for every
CVS operation, including making diffs.)

What I'd like to be able to do is (fairly) seamlessly edit my local
copy of egcs/wwwdocs stuff, then produce diffs while off-line.  (Emacs
20.3.1 doesn't produce backup versions in my local copy of the
repository, presumably because that doesn't make sense for systems
with live access to the repository.)  These diffs would be against
the earlier version I checked out from the repository -- not necessarily
against the most up-to-date versions, but that's normal.

That way, I can email the diffs to egcs-patches while off-line, batch
up a bunch at a time, and upload the emails en masse as part of my
normal email activity.

At the same time, it'd be great if I could batch egcs commits similarly,
orthagonally to producing patches.

For example, it'd be nice to be able to "prepare" a commit, including
the log message, and have that queued up for the next time I go on-line.
(Rejected commits would produce bounce-like messages emailed to me,
or something.)

Some of these things could probably be brute-forced just by making
multiple copies of the repository.

But, I'm sure I could work out and implement solutions using Emacs
lisp, shell scripts, and the like, just as I did (pretty well) for
my pre-CVS/pre-PPP setup.  That way I could save on disk space and
time (though it amazes me how fast my system can do a recursive
diff on two egcs trees, given how much stuff is in those...still,
it's longer than I want to wait to do ordinary transactions).

I'd rather find out what others have done in these directions first,
and see what use I can make of it.

Note that I haven't yet obtained or read any books on CVS, and most
of my recent experience with version-control systems involves using
ClearCase with on-line access to the repository (though some developers
had access similar to what I have now, so there were nightly merges,
etc.).  So, I think I understand the concepts and some of the pitfalls
pretty well, but I definitely do not know CVS and other specifics
all that well yet.  Recommendations for reading material are definitely
welcome.

        tq vm, (burley)

Re: Any recommendations for off-line CVS/egcs work?

[Dave Love]

>>>>> "JCB" == Craig Burley <craig@jcb-sc.com> writes:

 JCB> (Emacs 20.3.1 doesn't produce backup versions in my local copy
 JCB> of the repository,

You can customize `vc-make-backup-files'.  It is also possible to keep
local copies under RCS instead of using numeric backups, but possibly
more confusing than is worthwhile.

 JCB> So, I think I understand the concepts and some of the pitfalls
 JCB> pretty well, but I definitely do not know CVS and other
 JCB> specifics all that well yet.  Recommendations for reading
 JCB> material are definitely welcome.

In addition to the doc in the distribution, there's various material
at cyclic.com.  AFAIR it includes extra tutorial stuff and comparisons
with things like Clearcase.  Also AFAIR, it's not immediately obvious
where to find it all...

BTW, if people have suggestions for Emacs VC mode, now is a good time
to make them (to Spiegel, the maintainer).  There's a rewrite of VC
scheduled and the decks should be generally clear soon for other
non-bugfix Emacs changes.

Re: Any recommendations for off-line CVS/egcs work?

[Dave Love]

>>>>> "JCB" == Craig Burley <craig@jcb-sc.com> writes:

 JCB> (Emacs 20.3.1 doesn't produce backup versions in my local copy
 JCB> of the repository,

You can customize `vc-make-backup-files'.  It is also possible to keep
local copies under RCS instead of using numeric backups, but possibly
more confusing than is worthwhile.

 JCB> So, I think I understand the concepts and some of the pitfalls
 JCB> pretty well, but I definitely do not know CVS and other
 JCB> specifics all that well yet.  Recommendations for reading
 JCB> material are definitely welcome.

In addition to the doc in the distribution, there's various material
at cyclic.com.  AFAIR it includes extra tutorial stuff and comparisons
with things like Clearcase.  Also AFAIR, it's not immediately obvious
where to find it all...

BTW, if people have suggestions for Emacs VC mode, now is a good time
to make them (to Spiegel, the maintainer).  There's a rewrite of VC
scheduled and the decks should be generally clear soon for other
non-bugfix Emacs changes.

Any recommendations for off-line CVS/egcs work?

[craig]

If anyone has pointers to information on, or boilerplate/sample scripts,
Emacs lisp codes, easing doing egcs work (via CVS) largely off-line,
please drop me a note.

Specifically, I have dial-up access at an hourly charge.  (Actually,
I get the first 100 hours per month for a flat rate, so the cost isn't
so much the problem as the inconvenience of connecting for every
CVS operation, including making diffs.)

What I'd like to be able to do is (fairly) seamlessly edit my local
copy of egcs/wwwdocs stuff, then produce diffs while off-line.  (Emacs
20.3.1 doesn't produce backup versions in my local copy of the
repository, presumably because that doesn't make sense for systems
with live access to the repository.)  These diffs would be against
the earlier version I checked out from the repository -- not necessarily
against the most up-to-date versions, but that's normal.

That way, I can email the diffs to egcs-patches while off-line, batch
up a bunch at a time, and upload the emails en masse as part of my
normal email activity.

At the same time, it'd be great if I could batch egcs commits similarly,
orthagonally to producing patches.

For example, it'd be nice to be able to "prepare" a commit, including
the log message, and have that queued up for the next time I go on-line.
(Rejected commits would produce bounce-like messages emailed to me,
or something.)

Some of these things could probably be brute-forced just by making
multiple copies of the repository.

But, I'm sure I could work out and implement solutions using Emacs
lisp, shell scripts, and the like, just as I did (pretty well) for
my pre-CVS/pre-PPP setup.  That way I could save on disk space and
time (though it amazes me how fast my system can do a recursive
diff on two egcs trees, given how much stuff is in those...still,
it's longer than I want to wait to do ordinary transactions).

I'd rather find out what others have done in these directions first,
and see what use I can make of it.

Note that I haven't yet obtained or read any books on CVS, and most
of my recent experience with version-control systems involves using
ClearCase with on-line access to the repository (though some developers
had access similar to what I have now, so there were nightly merges,
etc.).  So, I think I understand the concepts and some of the pitfalls
pretty well, but I definitely do not know CVS and other specifics
all that well yet.  Recommendations for reading material are definitely
welcome.

        tq vm, (burley)