Gnus 5.11 in Emacs CVS

Gnus 5.11 in Emacs CVS

[Per Abrahamsen]

Anyone working on that?  

Judging by User-Agent stats, most people use the Gnus bundled with
Emacs, so I don't think we should consider any version of Gnus really
released until it is part of Emacs.

Re: Gnus 5.11 in Emacs CVS

[Steve Youngs]

[-- Attachment #1: Type: text/plain, Size: 926 bytes --]

|--==> "PA" == Per Abrahamsen <abraham@dina.kvl.dk> writes:

  PA> Judging by User-Agent stats, most people use the Gnus bundled
  PA> with Emacs,

I think it depends on which forums you hang out in.  Most of the ones I
see are either XEmacs package or Gnus.org CVS (or tarball).  Although
just looking at the User-Agent header doesn't tell me where they got
Gnus from.

  PA> so I don't think we should consider any version of Gnus really
  PA> released until it is part of Emacs.

I strongly disagree with that.  Any version of Gnus is released when
and only when Lars says it is released.  Gnus.org is the upstream for
the Gnus bundled with Emacs, not the other way around.

-- 
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
|              Ashes to ashes, dust to dust.               |
|      The proof of the pudding, is under the crust.       |
|------------------------------<sryoungs@bigpond.net.au>---|

[-- Attachment #2: Type: application/pgp-signature, Size: 256 bytes --]

Re: Gnus 5.11 in Emacs CVS

[Reiner Steib]

On Mon, Jan 05 2004, Steve Youngs wrote:

> |--==> "PA" == Per Abrahamsen <abraham@dina.kvl.dk> writes:
>
>   PA> Judging by User-Agent stats, most people use the Gnus bundled
>   PA> with Emacs,
>
> I think it depends on which forums you hang out in.  

Sure.  But many users hesitate or even refuse to update their Gnus,
even if the use ancient versions like 5.7 (with Emacs 20.7 for
example).

> Most of the ones I see are either XEmacs package or Gnus.org CVS (or
> tarball).

> Although just looking at the User-Agent header doesn't tell me where
> they got Gnus from.

For Emacs it's quite simple, isn't it?
Gnus 5.7, 5.9, 5.11 ==> bundled with Emacs.
Gnus 5.8.x, 5.10.x ==> Gnus.org.

Anyhow, Gnus 5.11 is ready for inclusion in Emacs CVS and this should
be done soon.  (How is the usual procedure for this?)

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo--- PGP key available via WWW   http://rsteib.home.pages.de/

Re: Gnus 5.11 in Emacs CVS

[Lars Magne Ingebrigtsen]

Reiner Steib <4.uce.03.r.s@nurfuerspam.de> writes:

> Anyhow, Gnus 5.11 is ready for inclusion in Emacs CVS and this should
> be done soon.  (How is the usual procedure for this?)

Some kind soul takes it upon him- or herself to start merging Gnus
into Emacs.  This is quite a bit of work.  

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: Gnus 5.11 in Emacs CVS

[Reiner Steib]

On Tue, Jan 06 2004, Lars Magne Ingebrigtsen wrote:

> Reiner Steib <4.uce.03.r.s@nurfuerspam.de> writes:
>
>> Anyhow, Gnus 5.11 is ready for inclusion in Emacs CVS and this should
>> be done soon.  (How is the usual procedure for this?)
>
> Some kind soul takes it upon him- or herself to start merging Gnus
> into Emacs.  This is quite a bit of work.  

Naively, I would have though that it's more or less placing
[gnus/]lisp files in [emacs]/lisp/gnus and [gnus/]texi in [emacs/]man.

Could someone who did (or helped with) the previous merge (Gnus 5.9 in
Emacs 21) tell a little more about it?  Possible problems, pitfalls,
etc.?

What about all the changes made to Gnus 5.8.8/5.9 [1] in Emacs' CVS?
Probably the changes committed by Dave Love, Jesper Harder, ShengHuo
ZHU or other developers with write access to Gnus' CVS were also
applied to (or synced with) Gnus' CVS.  I didn't check, though.

Bye, Reiner.

[1] It's not clear to me from the ChangeLog (in Emacs CVS), when this
    merge happened.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo--- PGP key available via WWW   http://rsteib.home.pages.de/

Re: Gnus 5.11 in Emacs CVS

[Reiner Steib]

On Thu, Jan 08 2004, Reiner Steib wrote:

> On Tue, Jan 06 2004, Lars Magne Ingebrigtsen wrote:
[ Including Gnus 5.11 in Emacs CVS ]
>> Some kind soul takes it upon him- or herself to start merging Gnus
>> into Emacs.  This is quite a bit of work.  
>
> Naively, I would have though that it's more or less placing
> [gnus/]lisp files in [emacs]/lisp/gnus and [gnus/]texi in [emacs/]man.

I gave it a try and it seems to work nicely.  I wrote down the
necessary steps as a shell script (see below), that should be run in
directory "emacs/" of an Emacs check-out before "make bootstrap".
Maybe someone wants to test it too.

Some minor problems (see the comments about "load-path shadows" in the
script):

# Not needed for GNU Emacs (XEmacs compatibility library)
  $rm gnus/gnus-xmas.el{,c}

# Gnus can use the file from calendar/ instead:
  $rm gnus/parse-time.el{,c}

# Gnus can use the file from net/ instead:
  $rm gnus/netrc.el{,c}

The following items should probably be discussed on emacs-devel:

# We should probably move gnus/tls.el to net/
  $rm net/tls.el{,c}

# The version from calendar/ lacks `time-to-number-of-days'
  $rm calendar/time-date.el{,c}

Other items: 

- What about GNUS-NEWS -> etc/NEWS?  Where to put the Gnus items?
  `**' has to be replaced by `***'.  Should the item about "make.bat"
  be removed (it doesn't make sense for the bundled Gnus, I suppose)?

- What about the arch-tags in Emacs CVS?  Probably we should ask Miles
  Bader (who added them) for advice (on emacs-devel)?

I still would like to hear comments on the following:

> Could someone who did (or helped with) the previous merge (Gnus 5.9 in
> Emacs 21) tell a little more about it?  Possible problems, pitfalls,
> etc.?
>
> What about all the changes made to Gnus 5.8.8/5.9 [1] in Emacs' CVS?
> Probably the changes committed by Dave Love, Jesper Harder, ShengHuo
> ZHU or other developers with write access to Gnus' CVS were also
> applied to (or synced with) Gnus' CVS.  I didn't check, though.
[...]
> [1] It's not clear to me from the ChangeLog (in Emacs CVS), when this
>     merge happened.

Bye, Reiner.

--8<---------------cut here---------------start------------->8---
#!/usr/bin/env bash
# With GNU cp, rm, ...
: ${rm:="rm -v"}
: ${cp:="cp -vp"}

old_pwd=`pwd`

# Base directory of your unpacked Gnus 5.10.6 source tarball
# (must have the subdirs etc/, lisp/ and texi/):
: ${Oort_base:="$HOME/src/links/Emacs/gnus/5.10/gnus-5.10.6"}

dirs_to_check="$Oort_base/texi
               $Oort_base/etc
	       $Oort_base/lisp
	       lisp/gnus
	       man"

for i in $dirs_to_check; do     
  [ -d "$i" ] || {
    echo "'$i' is not a directory."
    echo "Is '$old_pwd' (pwd)"
    echo "really the base directory of your emacs checkout"
    echo "and is '$Oort_base' the correct Gnus 5.10.6 directory?"
    exit 1
  } >&2
done

manuals="emacs-mime.texi gnus-faq.texi gnus.texi message.texi
         pgg.texi sieve.texi"

(
  echo "Setting up 'man'..." >&2
  cd man
  echo "Removing old *.texi files..." >&2
  for i in $manuals; do
    $rm $i
    $cp $Oort_base/texi/$i $i
  done
)

(
  echo "Setting up 'lisp/gnus'..." >&2
  cd lisp/gnus
  echo "Removing old *.el and *.?pm files..." >&2
  $rm *.el *.elc *.?pm
  echo "New pixmaps..."  >&2
  $cp $Oort_base/etc/{smilies,gnus}/*.?pm .
  echo "New lisp files..."  >&2
  $cp $Oort_base/lisp/*.el .
)

(
  cd lisp
  diff_files="net/tls.el gnus/tls.el
              net/netrc.el gnus/netrc.el
              calendar/time-date.el gnus/time-date.el"
  
  (
    set -- $diff_files
    while [ $# -ge 2 ]; do
      diff -u $1 $2
      shift 2
    done > $old_pwd/Gnus-5-11.patch
  )

  (
    set -- $diff_files
    while [ $# -ge 2 ]; do
      diff -wu $1 $2
      shift 2
    done > $old_pwd/Gnus-5-11.no-ws.patch
  )

# Make sure that we don't have any load-path shadows:

  echo "removing some files.." >&2
# Not needed for GNU Emacs (XEmacs compatibility library)
  $rm gnus/gnus-xmas.el{,c}

# Gnus can use the file from calendar/ instead:
  $rm gnus/parse-time.el{,c}
# Gnus can use the file from net/ instead:
  $rm gnus/netrc.el{,c}

# The following should probably be discussed on emacs-devel:

# We should probably move gnus/tls.el to net/
  $rm net/tls.el{,c}
# The version from calendar/ lacks some function
  $rm calendar/time-date.el{,c}
)

exit
--8<---------------cut here---------------end--------------->8---

Re: Gnus 5.11 in Emacs CVS

[Jesper Harder]

Reiner Steib <4.uce.03.r.s@nurfuerspam.de> writes:

> I still would like to hear comments on the following:
>
>> What about all the changes made to Gnus 5.8.8/5.9 [1] in Emacs' CVS?
>> Probably the changes committed by Dave Love, Jesper Harder, ShengHuo
>> ZHU or other developers with write access to Gnus' CVS were also
>> applied to (or synced with) Gnus' CVS.  I didn't check, though.

Yes, all my changes are in Gnus CVS.

Re: Gnus 5.11 in Emacs CVS

[Jesper Harder]

Reiner Steib <4.uce.03.r.s@nurfuerspam.de> writes:

> Some minor problems (see the comments about "load-path shadows" in the
> script):
>
> # Not needed for GNU Emacs (XEmacs compatibility library)
>   $rm gnus/gnus-xmas.el{,c}

Also messagexmas.el, nnheaderxm.el, dgnushack.el gnus-load.el and
lpath.el.

> # Gnus can use the file from net/ instead:
>   $rm gnus/netrc.el{,c}

I think netrc.el is primarily maintained in Gnus, so the Gnus version
should be used.

> # The version from calendar/ lacks `time-to-number-of-days'
>   $rm calendar/time-date.el{,c}

Ditto.

> manuals="emacs-mime.texi gnus-faq.texi gnus.texi message.texi
>          pgg.texi sieve.texi"

We should remove the @iflatex stuff from gnus.texi

    sed -e '/@iflatex/,/@end iflatex/d' $< >

otherwise you can't use the normal procedure to tex it.

Other stuff to consider:

* The refcard.  emacs/etc already has refcards for dired, viper
  etc. so we might as well include it.

* The ChangeLogs.

* Fix some of the numerous byte-compiler warnings you get when you
  compile Gnus as a part of Emacs [1].

* Maybe some other files from Gnus should go to net/, e.g. dig.el,
  dns.el ... -- I don't know.

[1] In some ways I think we're doing ourselves a disservice with the
    elaborate compilation scheme.  With cvs Emacs (at least) most of
    it should be unnecessary because `(if (fboundp' kills the
    warnings.

Re: Gnus 5.11 in Emacs CVS

[Simon Josefsson]

Jesper Harder <harder@ifa.au.dk> writes:

> [1] In some ways I think we're doing ourselves a disservice with the
>     elaborate compilation scheme.  With cvs Emacs (at least) most of
>     it should be unnecessary because `(if (fboundp' kills the
>     warnings.

We could drop lpath/dgnushack in favor of automake's elisp support.  I
did that locally a few weeks ago, without any obvious problem.  It
generated a bunch of new byte compiler warnings, which we should fix.
I didn't handle the XEmacs specific part, but they could easily be
hidden in an automake condition, enabled by --with-xemacs or
emacs/xemacs guessing logic.

Opinions?  I guess Lars will have to decide, since there are several
rules in the makefiles that seem to be used by him only.

Re: Gnus 5.11 in Emacs CVS

[Jesper Harder]

Simon Josefsson <jas@extundo.com> writes:

> We could drop lpath/dgnushack in favor of automake's elisp support.
> I did that locally a few weeks ago, without any obvious problem.  It
> generated a bunch of new byte compiler warnings, which we should
> fix.

Regarding those warnings: Is there a reason that we often do

  (eval-when-compile
     (require 'foo))

rather than[1]

  (autoload 'foo-open "foo")

The later would remove the "might not be defined at runtime" warnings.


[1] or autoload cookies, if it's used in many files.

Re: Gnus 5.11 in Emacs CVS

[Simon Josefsson]

Jesper Harder <harder@ifa.au.dk> writes:

> Simon Josefsson <jas@extundo.com> writes:
>
>> We could drop lpath/dgnushack in favor of automake's elisp support.
>> I did that locally a few weeks ago, without any obvious problem.  It
>> generated a bunch of new byte compiler warnings, which we should
>> fix.
>
> Regarding those warnings: Is there a reason that we often do
>
>   (eval-when-compile
>      (require 'foo))
>
> rather than[1]
>
>   (autoload 'foo-open "foo")
>
> The later would remove the "might not be defined at runtime" warnings.
>
>
> [1] or autoload cookies, if it's used in many files.

I dunno.  For macros maybe?  The latter seems cleaner, though.

Re: Gnus 5.11 in Emacs CVS

[Reiner Steib]

On Wed, Jan 14 2004, Jesper Harder wrote:

> Reiner Steib <4.uce.03.r.s@nurfuerspam.de> writes:
[...]
>> # Not needed for GNU Emacs (XEmacs compatibility library)
>>   $rm gnus/gnus-xmas.el{,c}
>
> Also messagexmas.el, nnheaderxm.el, dgnushack.el gnus-load.el and
> lpath.el.

Yes.

>> # Gnus can use the file from net/ instead:
>>   $rm gnus/netrc.el{,c}
>
> I think netrc.el is primarily maintained in Gnus, so the Gnus version
> should be used.

I took the version from net/ because the changes [1] there seem to be
reasonable.  If there has been a decision to move the file from gnus/
to net/ in the past, it's probably not worth to discuss this again,
unless there are new facts to consider, IMHO.

[ Some related discussions:

- "using qp.el, rfc2047.el in mailutils, rmail"
  <URL:http://thread.gmane.org/gmane.emacs.pretest.bugs/1459>
  <news:E1AZtzf-0007c8-J6@fencepost.gnu.org>

- "Merging quoted-printable functions"
  <URL:http://article.gmane.org/gmane.emacs.devel/18882>
  <news:E1AaMjp-0004fu-VM@fencepost.gnu.org> ]

>> # The version from calendar/ lacks `time-to-number-of-days'
>>   $rm calendar/time-date.el{,c}
>
> Ditto.

Yes.

> We should remove the @iflatex stuff from gnus.texi
>
>     sed -e '/@iflatex/,/@end iflatex/d' $< >
>
> otherwise you can't use the normal procedure to tex it.

"make dvi" gave me a nice "gnus.dvi".  Ah, because of this:

,----
| gnus.dvi: gnus.texi
|         sed -e '/@iflatex/,/@end iflatex/d' ${srcdir}/gnus.texi > gnustmp.texi
|         $(ENVADD) $(TEXI2DVI) gnustmp.texi
|         cp gnustmp.dvi $*.dvi
|         rm gnustmp.*
`----

But we have to add "sieve" and "pgg" to dvi and info targets (as we
have for "message").  See [2].

> Other stuff to consider:
>
> * The refcard.  emacs/etc already has refcards for dired, viper
>   etc. so we might as well include it.

Sure.

> * The ChangeLogs.

Yes, but I don't know what should happen to the entries from
"5.9"-inclusion until today in lisp/gnus/ChangeLog.

> * Maybe some other files from Gnus should go to net/, e.g. dig.el,
>   dns.el ... -- I don't know.

This could be discussed (and done) later, I guess.

Bye, Reiner.

,----[1]
| --- emacs/lisp/net/netrc.el     2003-09-01 17:45:33.000000000 +0200
| +++ gnus-5.10.6/lisp/netrc.el   2002-04-23 19:09:46.000000000 +0200
| @@ -4,7 +4,6 @@
|  
|  ;; Author: Lars Magne Ingebrigtsen <larsi@gnus.org>
| +;; Modularizer: Ted Zlatanov <tzz@lifelogs.com>
|  ;; Keywords: news
| -;;  Modularized by Ted Zlatanov <tzz@lifelogs.com>
| -;;  when it was part of Gnus.
|  
|  ;; This file is part of GNU Emacs.
| @@ -36,11 +35,12 @@
|  ;;;
|  
| +(eval-and-compile
|  (defalias 'netrc-point-at-eol
|    (if (fboundp 'point-at-eol)
|        'point-at-eol
| -    'line-end-position))
| +      'line-end-position)))
|  
|  (defun netrc-parse (file)
| -  "Parse FILE and return a list of all entries in the file."
| +  "Parse FILE and return an list of all entries in the file."
|    (when (file-exists-p file)
|      (with-temp-buffer
| @@ -126,4 +126,3 @@
|  (provide 'netrc)
|  
| -;;; arch-tag: af9929cc-2d12-482f-936e-eb4366f9fa55
|  ;;; netrc.el ends here
`----

[2] Changes to man/Makefile.in:
--8<---------------cut here---------------start------------->8---
--- Makefile.in.~1.52.~	Wed Oct 22 10:30:13 2003
+++ Makefile.in	Wed Jan 14 15:14:57 2004
@@ -34,18 +34,21 @@
 # The makeinfo program is part of the Texinfo distribution.
 MAKEINFO = makeinfo
 INFO_TARGETS = ../info/emacs ../info/ccmode ../info/cl \
-	        ../info/dired-x ../info/ediff ../info/forms ../info/gnus \
-		../info/info ../info/message ../info/mh-e ../info/reftex \
+	        ../info/dired-x ../info/ediff ../info/forms ../info/info \
+                ../info/gnus ../info/message ../info/sieve ../info/pgg \
+                ../info/emacs-mime ../info/mh-e ../info/reftex \
 		../info/sc ../info/vip ../info/viper ../info/widget \
 		../info/efaq ../info/ada-mode ../info/autotype ../info/calc \
 		../info/idlwave ../info/eudc ../info/ebrowse ../info/pcl-cvs \
-		../info/woman ../info/emacs-mime ../info/eshell \
+		../info/woman ../info/eshell \
 		../info/speedbar ../info/tramp ../info/ses ../info/smtpmail
 DVI_TARGETS = 	emacs.dvi calc.dvi cc-mode.dvi cl.dvi dired-x.dvi \
-		 ediff.dvi forms.dvi gnus.dvi message.dvi mh-e.dvi \
+		 ediff.dvi forms.dvi \
+                 gnus.dvi message.dvi sieve.dvi pgg.dvi emacs-mime.dvi \
+                 mh-e.dvi \
 		 reftex.dvi sc.dvi vip.dvi viper.dvi widget.dvi faq.dvi \
 		 ada-mode.dvi autotype.dvi idlwave.dvi eudc.dvi ebrowse.dvi \
-		 pcl-cvs.dvi woman.dvi emacs-mime.dvi eshell.dvi \
+		 pcl-cvs.dvi woman.dvi eshell.dvi \
 		 speedbar.dvi tramp.dvi ses.dvi smtpmail.dvi
 INFOSOURCES = info.texi
 
@@ -185,11 +188,18 @@
 	$(ENVADD) $(TEXI2DVI) gnustmp.texi
 	cp gnustmp.dvi $*.dvi
 	rm gnustmp.*
-
 ../info/message: message.texi
 	cd $(srcdir); $(MAKEINFO) message.texi
 message.dvi: message.texi
 	$(ENVADD) $(TEXI2DVI) ${srcdir}/message.texi
+../info/sieve: sieve.texi
+	cd $(srcdir); $(MAKEINFO) sieve.texi
+sieve.dvi: sieve.texi
+	$(ENVADD) $(TEXI2DVI) ${srcdir}/sieve.texi
+../info/pgg: pgg.texi
+	cd $(srcdir); $(MAKEINFO) pgg.texi
+pgg.dvi: pgg.texi
+	$(ENVADD) $(TEXI2DVI) ${srcdir}/pgg.texi
 
 ../info/mh-e: mh-e.texi
 	cd $(srcdir); $(MAKEINFO) mh-e.texi
--8<---------------cut here---------------end--------------->8---
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo--- PGP key available via WWW   http://rsteib.home.pages.de/

Re: Gnus 5.11 in Emacs CVS

[Jesper Harder]

Reiner Steib <4.uce.03.r.s@nurfuerspam.de> writes:

> On Thu, Jan 08 2004, Reiner Steib wrote:
>
>> On Tue, Jan 06 2004, Lars Magne Ingebrigtsen wrote:
>
>>> Some kind soul takes it upon him- or herself to start merging Gnus
>>> into Emacs.  This is quite a bit of work.  
>>
>> Naively, I would have though that it's more or less placing
>> [gnus/]lisp files in [emacs]/lisp/gnus and [gnus/]texi in
>> [emacs/]man.

Maybe it would make sense to make a branch in Gnus cvs and start doing
the necessary stuff, deleting files, merging from Emacs etc. there?

Re: Gnus 5.11 in Emacs CVS

[Reiner Steib]

On Thu, Jan 22 2004, Jesper Harder wrote:

>> On Thu, Jan 08 2004, Reiner Steib wrote:
[...]
>>> Naively, I would have though that it's more or less placing
>>> [gnus/]lisp files in [emacs]/lisp/gnus and [gnus/]texi in
>>> [emacs/]man.
>
> Maybe it would make sense to make a branch in Gnus cvs and start doing
> the necessary stuff, deleting files, merging from Emacs etc. there?

If we need a branch, shouldn't it be in _Emacs_ CVS to test the
integration (e.g. check that new files have to be added to CVS, ...)?
I wonder how the merge has been done last time.  I'd like to help with
it, but probably it would make more sense if someone with write access
to Emacs CVS does the first (big) step.

Maybe we should shift (or cross-post) this discussion to emacs-devel?

BTW, I've added the suggestions from your previous article [1] to my
Emacs-CVS checkout [2] and I'm using it for more than a week without
problems.

Bye, Reiner.

[1] See <news:m3u12z9klj.fsf@defun.localdomain> and
    <news:v94quywr48.fsf@marauder.physik.uni-ulm.de>.
    Not addressed yet: byte-compiler warnings

[2] The updated shell script is available here:
    <URL:http://theotp1.physik.uni-ulm.de/~ste/comp/emacs/gnus/Gnus-5-11.sh>
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo--- PGP key available via WWW   http://rsteib.home.pages.de/

Re: Gnus 5.11 in Emacs CVS

[Miles Bader]

Reiner Steib <4.uce.03.r.s@nurfuerspam.de> writes:
> If we need a branch, shouldn't it be in _Emacs_ CVS to test the
> integration (e.g. check that new files have to be added to CVS, ...)?
> I wonder how the merge has been done last time.  I'd like to help with
> it, but probably it would make more sense if someone with write access
> to Emacs CVS does the first (big) step.

Instead of CVS, you could use tla (aka `arch') and my emacs arch branch,
which is synchronized with the emacs CVS trunk.

tla allows (painlessly) distributed repositories, and generally uses a
`pull' model, so write access to a central repository isn't necessary.
It is also much better at merging than CVS.

***
One part of this is that I've added `taglines' to all the emacs source
files (in CVS), which make it easy to assign a unique identity to each
file (even if the name changes).

If you'd accept a patch against Gnus CVS to add taglines, this would
make future merging using tla easier.

This is slightly trickier than with a new project, because Gnus and
Emacs conceptually _share_ files (even though the actual contents may
vary), so it would be best to use the _same_ tagline for the same file
in both Gnus and Emacs.

An example of a tagline is something like `lisp/gnus/gnus-art.el' from
the emacs source tree:

   ....
   (run-hooks 'gnus-art-load-hook)

   ;;; arch-tag: 2654516f-6279-48f9-a83b-05c1fa450c33
   ;;; gnus-art.el ends here


> Maybe we should shift (or cross-post) this discussion to emacs-devel?

Couldn't hurt... [note CC: :-]

-Miles
-- 
`Suppose Korea goes to the World Cup final against Japan and wins,' Moon said.
`All the past could be forgiven.'   [NYT]

Re: Gnus 5.11 in Emacs CVS

[Miles Bader]

[-- Attachment #1: Type: text/plain, Size: 425 bytes --]

Miles Bader <miles@gnu.org> writes:
> Instead of CVS, you could use tla (aka `arch') and my emacs arch branch,
> which is synchronized with the emacs CVS trunk.

Oh, BTW, for anyone who already uses tla the branch is:

   miles@gnu.org--gnu-2004/emacs--cvs-trunk--0

The archive `miles@gnu.org--gnu-2004' is available at:

   http://sourcecontrol.net/~miles/miles@gnu.org--gnu-2004

The archive is signed with my GPG keys:



[-- Attachment #2: GPG keys for miles@gnu.org --]
[-- Type: application/pgp-keys, Size: 3883 bytes --]

[-- Attachment #3: Type: text/plain, Size: 91 bytes --]





-Miles
-- 
.Numeric stability is probably not all that important when you're guessing.

Re: Gnus 5.11 in Emacs CVS

[Reiner Steib]

On Sun, Feb 15 2004, Miles Bader wrote:

> Reiner Steib <4.uce.03.r.s@nurfuerspam.de> writes:
>> If we need a branch, shouldn't it be in _Emacs_ CVS to test the
>> integration (e.g. check that new files have to be added to CVS, ...)?
>> I wonder how the merge has been done last time.  I'd like to help with
>> it, but probably it would make more sense if someone with write access
>> to Emacs CVS does the first (big) step.
>
> Instead of CVS, you could use tla (aka `arch') and my emacs arch branch,
> which is synchronized with the emacs CVS trunk.
>
> tla allows (painlessly) distributed repositories, and generally uses a
> `pull' model, so write access to a central repository isn't necessary.
> It is also much better at merging than CVS.

Do I understand you correctly that if I managed to merge Gnus into an
arch archive of Emacs, you would sync the archive to Emacs CVS trunk?

I've never used tla before.  I installed it now and will read some
introductory documentation soon (from wiki.gnuarch.org).  It would be
very helpful for me if you could give me a brief (or even better: not
so brief ;-)) description of the necessary steps.

> One part of this is that I've added `taglines' to all the emacs source
> files (in CVS), which make it easy to assign a unique identity to each
> file (even if the name changes).
>
> If you'd accept a patch against Gnus CVS to add taglines, this would
> make future merging using tla easier.

I cannot speak for Lars, but I don't think he would refuse.  Lars?
(If Lars agrees on this, I can commit your patches to Gnus CVS).

> This is slightly trickier than with a new project, because Gnus and
> Emacs conceptually _share_ files (even though the actual contents may
> vary), so it would be best to use the _same_ tagline for the same file
> in both Gnus and Emacs.

Apart from the trunk, we have a branch "v5-10" for the 5.10 series.
Probably they should be added in both branches?

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo--- PGP key available via WWW   http://rsteib.home.pages.de/

Re: Gnus 5.11 in Emacs CVS

[Miles Bader]

Reiner Steib <4.uce.03.r.s@nurfuerspam.de> writes:
> > Instead of CVS, you could use tla (aka `arch') and my emacs arch branch,
> > which is synchronized with the emacs CVS trunk.
> >
> > tla allows (painlessly) distributed repositories, and generally uses a
> > `pull' model, so write access to a central repository isn't necessary.
> > It is also much better at merging than CVS.
> 
> Do I understand you correctly that if I managed to merge Gnus into an
> arch archive of Emacs, you would sync the archive to Emacs CVS trunk?

Well, technically I can do it -- of course I'd have to establish that it's
OK with the maintainers; if it's agreed that you're an authoritative source
for Gnus changes in the emacs tree, then it should be very easy to do this
in the future.

> I've never used tla before.  I installed it now and will read some
> introductory documentation soon (from wiki.gnuarch.org).  It would be
> very helpful for me if you could give me a brief (or even better: not
> so brief ;-)) description of the necessary steps.

Er, it depends on what you want to do.  For simply getting a tree, it's
pretty easy.  Here's some steps to do so from scratch (including some
optional setup that should make things work better):


##### Once-only first-time setup commands:

   tla my-id ...

# For efficiency, it's also recommended to establish a `revision library'
# on the local hard disk; this can speed up operation quit a bit

   mkdir -p /usr/local/arch/revlib
   tla my-revision-library /usr/local/arch/revlib
   tla library-config --greedy --sparse /usr/local/arch/revlib

# To verify signed archives, you probably want to do this:

   mkdir -p ~/.arch-params/signing
   echo 'gpg --verify-files -' > ~/.arch-params/signing/=default.check

   # [make sure you have keys for anyone you're interested in -- my keys
   #  are available at pgp.mit.edu as "miles@gnu.org"]


##### Once for any new archive you use:

# You need to do this once for any new archive you use:

   tla register-archive http://sourcecontrol.net/~miles/miles@gnu.org--gnu-2004


##### Regular commands:

# Checkout emacs into a directory called `/usr/local/src/emacs':

   cd /usr/local/src
   tla get emacs--cvs-trunk--0 emacs

# ... eventually, if you want to update your check-out to the latest version:

   cd /usr/local/src/emacs
   tla update


[For doing development using arch, you need more commands of course.]

For getting a more detailed view of how arch works, the `arch tutorial' is
very good (I think there should be an obvious pointer on the wiki?).

> > If you'd accept a patch against Gnus CVS to add taglines, this would
> > make future merging using tla easier.
> 
> I cannot speak for Lars, but I don't think he would refuse.  Lars?
> (If Lars agrees on this, I can commit your patches to Gnus CVS).
> 
> > This is slightly trickier than with a new project, because Gnus and
> > Emacs conceptually _share_ files (even though the actual contents may
> > vary), so it would be best to use the _same_ tagline for the same file
> > in both Gnus and Emacs.
> 
> Apart from the trunk, we have a branch "v5-10" for the 5.10 series.
> Probably they should be added in both branches?

What might be easiest is for you to merge the Gnus changes from the
emacs tree back into the canonical Gnus sources (including the branch).
As the emacs Gnus files all have had taglines added, this will have the
effect of adding them to the Gnus sources too.

Any file not shared with emacs will also need to have taglines added, but
that's easy to do; I'm mainly concerned that the common files end up with
the same taglines in the Gnus and Emacs trees.

At that point, I'd like to make arch branches for Gnus, synchronized with Gnus
CVS (just one-way of course -- unless you want to give me write access :-),
probably a few, e.g., `gnus--cvs-trunk--0', and `gnus--rel--5.10'.

Thanks,

-Miles
-- 
 ___
(^o^;
(()))
*This is the cute octopus virus, please copy it into your sig so it can spread.

Re: Gnus 5.11 in Emacs CVS

[Reiner Steib]

On Thu, Apr 15 2004, Miles Bader wrote:

> Reiner Steib <4.uce.03.r.s@nurfuerspam.de> writes:
[...]
>> Do I understand you correctly that if I managed to merge Gnus into an
>> arch archive of Emacs, you would sync the archive to Emacs CVS trunk?
>
> Well, technically I can do it -- of course I'd have to establish that it's
> OK with the maintainers; [...]

Just a pointer:

The Emacs maintainers seem to agree that Gnus 5.10 (or 5.11 after
renaming) can be merged into the current Emacs trunk (which is in
feature freeze for the next Emacs release).  See
<URL:http://thread.gmane.org/gmane.emacs.devel/22994>.

Further discussions on the merge should be done on `emacs-devel'.
(Cc-ing the Gnus list is okay, of course.)

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo--- PGP key available via WWW   http://rsteib.home.pages.de/

Re: Gnus 5.11 in Emacs CVS

[Xavier Maillard]

[-- Attachment #1: Type: text/plain, Size: 884 bytes --]

On 5 Jan 2004, Steve Youngs<sryoungs@bigpond.net.au> stated:
>> --==> "PA" == Per Abrahamsen <abraham@dina.kvl.dk> writes:
>
> PA> Judging by User-Agent stats, most people use the Gnus bundled
> PA> with Emacs,
>
> I think it depends on which forums you hang out in. Most of the ones I
> see are either XEmacs package or Gnus.org CVS (or tarball). Although
> just looking at the User-Agent header doesn't tell me where they got
> Gnus from.
>
> PA> so I don't think we should consider any version of Gnus really
> PA> released until it is part of Emacs.
>
> I strongly disagree with that.  Any version of Gnus is released when
> and only when Lars says it is released.  Gnus.org is the upstream for
> the Gnus bundled with Emacs, not the other way around.

I totally agree with you, Lars is THE ONE who says a new Gnus is
released.

zeDek
-- 
.o.  | Hacker wonderland
..o  |
ooo  | 


[-- Attachment #2: Type: application/pgp-signature, Size: 188 bytes --]

Re: Gnus 5.11 in Emacs CVS

[Per Abrahamsen]

Steve Youngs <sryoungs@bigpond.net.au> writes:

> I think it depends on which forums you hang out in.  

I ran some stats on the all groups in the newsserver, but they are
pretty dated.  Back then the bundled version was more popular than the
rest combined.

> Although
> just looking at the User-Agent header doesn't tell me where they got
> Gnus from.

The version bundled with Emacs tend to have a unique version number,
different from the XEmacs packages and Lars' distribution.

> Gnus.org is the upstream for the Gnus bundled with Emacs, not the
> other way around.

That was not the question.

Basically, I consider getting the software out where non-enthusiasts
use it part of the release process.  Otherwise, the project risks
getting too inbreed, with a few people writing for each other.