Song of Programming

Song of Programming

[Lars Magne Ingebrigtsen]

So, I'm taking a week off work (I hope) now, and I'm releasing Gnus
5.8.0 at the end of it (I hope).  A few more Pterodactyls first,
though.

Before that, I'll be answering mail and stuff, but are there any
particular outstanding issues that should be resolved before going
beta and freezing features?  (Except fixing the usual bugs?)

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

Re: Song of Programming

[Hrvoje Niksic]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Before that, I'll be answering mail and stuff, but are there any
> particular outstanding issues that should be resolved before going
> beta and freezing features?

Oh yes: the variables that control and customize the display of the
MIME parts.  When asked about this several times, you said you'd
revamp that.  It is probably a very good idea to do so.

Ideally, it should be easy to specify a custom function for displaying 
a MIME part, or a set thereof.

Re: Song of Programming

[Hrvoje Niksic]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> So, I'm taking a week off work (I hope) now, and I'm releasing Gnus
> 5.8.0 at the end of it (I hope).  A few more Pterodactyls first,
> though.
> 
> Before that, I'll be answering mail and stuff, but are there any
> particular outstanding issues that should be resolved before going
> beta and freezing features?

Also: wasn't IMAP support supposed to be integrated?  Or is that too
unfinished?

Re: Song of Programming

[William M. Perry]

Lars Magne Ingebrigtsen wrote:

> So, I'm taking a week off work (I hope) now, and I'm releasing Gnus
> 5.8.0 at the end of it (I hope).  A few more Pterodactyls first,
> though.
>
> Before that, I'll be answering mail and stuff, but are there any
> particular outstanding issues that should be resolved before going
> beta and freezing features?  (Except fixing the usual bugs?)

Splitting into multiple backends would earn you my undying gratitude.
:)  It is that or I will be forced to use Exchange to set up the
auto-splitting rules on the IMAP server every once in a while, we I do
_not_ want to have to do. :(

-Bill P.

Re: Song of Programming

[Per Abrahamsen]

"William M. Perry" <wmperry@aventail.com> writes:

> Splitting into multiple backends would earn you my undying gratitude.
> :)  

That's new functionality.  I really think we should get the MIME
version out as fast as possible, it has really become a *must*,
especially for mail.

The only thing I'm not sure of, is if the MIME API's are ready.
Are they? 

Re: Song of Programming

[Harry Putnam]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> So, I'm taking a week off work (I hope) now, and I'm releasing Gnus
> 5.8.0 at the end of it (I hope).  A few more Pterodactyls first,
> though.
> 
> Before that, I'll be answering mail and stuff, but are there any
> particular outstanding issues that should be resolved before going
> beta and freezing features?  (Except fixing the usual bugs?)


The fact that gnus-agent looses track of agentized posts once they are
expunged from the server is a singularly annoying aspect of gnus, for
us offliners.

Warning messages that bring up a choice [y or n], to go on-line for
actions that require onlinedness, such as "g" in group buffer "F" in
group buffer etc.  This would prevent the occurrence or another bug
that can remove readedness marks from agentized groups in certain
conditions involving getting news with "g" when unplugged.

Re: Song of Programming

[Alexandre Oliva]

On Jul  3, 1999, Lars Magne Ingebrigtsen <larsi@gnus.org> wrote:

> Before that, I'll be answering mail and stuff, but are there any
> particular outstanding issues that should be resolved before going
> beta and freezing features?  (Except fixing the usual bugs?)

I'd really appreciate to have the fancy splitting patch I posted
installed in the next release, as well as the group-params based mail
splitting mechanism.  You had said the latter would have gone in 5.6.0
if I had already had my copyright assignment on file.  Now it's been
on file for a while. :-)

-- 
Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il
{oliva,Alexandre.Oliva}@dcc.unicamp.br  aoliva@{acm.org,computer.org}
oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org}
*** E-mail about software projects will be forwarded to mailing lists

Re: Song of Programming

[Per Abrahamsen]

Alexandre Oliva <oliva@dcc.unicamp.br> writes:

> I'd really appreciate to have the fancy splitting patch I posted
> installed in the next release, as well as the group-params based mail
> splitting mechanism.  

I'd like those kind of changes to have a longer alpha testing period,
Errors in the mail splitting code can easily lead to lost mail, even
in ways that aren't discovered for a while.

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Hrvoje Niksic <hniksic@srce.hr> writes:

> Oh yes: the variables that control and customize the display of the
> MIME parts.  When asked about this several times, you said you'd
> revamp that.  It is probably a very good idea to do so.

Yes.  I'm growing a bit blind in this area, though -- the current
state of affairs is growing more and more reasonable in my eyes.  :-) 
Perhaps somebody could have a gander at it and suggest a cleanup?

> Ideally, it should be easy to specify a custom function for displaying 
> a MIME part, or a set thereof.

Yes.

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

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Hrvoje Niksic <hniksic@srce.hr> writes:

> Also: wasn't IMAP support supposed to be integrated?  Or is that too
> unfinished?

The papers are underway, I think, and it'll be included this week.
Probably. 

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

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Per Abrahamsen <abraham@dina.kvl.dk> writes:

> The only thing I'm not sure of, is if the MIME API's are ready.
> Are they? 

I retract my statement about the MIME variables being reasonable.  I'm 
trying to write a section documenting them, and I have no idea what
all these variables do.  :-(

It's clean-up-time!

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

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Harry Putnam <reader@newsguy.com> writes:

> The fact that gnus-agent looses track of agentized posts once they are
> expunged from the server is a singularly annoying aspect of gnus, for
> us offliners.

Yup.  Fix in Pterodactyl Gnus v0.91.

> Warning messages that bring up a choice [y or n], to go on-line for
> actions that require onlinedness, such as "g" in group buffer "F" in
> group buffer etc.

Well, `F' and `g' are completely valid when unplugged.

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

Re: Song of Programming

[Harry Putnam]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Harry Putnam <reader@newsguy.com> writes:
> 
> > The fact that gnus-agent looses track of agentized posts once they are
> > expunged from the server is a singularly annoying aspect of gnus, for
> > us offliners.
> 
> Yup.  Fix in Pterodactyl Gnus v0.91.

Oh goody, ... That'll be a nice addition.


> > Warning messages that bring up a choice [y or n], to go on-line for
> > actions that require onlinedness, such as "g" in group buffer "F" in
> > group buffer etc.
> 
> Well, `F' and `g' are completely valid when unplugged.

OOPs ... F does do something usefull when unplugged, but what does "g"
do when unplugged.  All I ever see is a message "checking new news
.. done" but nothing actually happens, except occasionally it causes
one or several agentized nntp groups to loose all track of readedness,
so that when pressing "g" while next plugged, all messages on the
server show up as unread.  Not consistenly and usually just one
certain group, so probably something local.  Haven't nailed it down.

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Harry Putnam <reader@newsguy.com> writes:

> OOPs ... F does do something usefull when unplugged, but what does "g"
> do when unplugged.

The same as when it's plugged -- it reads the active file and stuff.
Of course, if you only update the Agent active file from the same Gnus 
that you're reading unplugged, there won't be anything new in there,
but I use a separate Gnus to fetch news/mail from the one I read in.
The one I read news in is always unplugged.

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

Re: Song of Programming

[Harry Putnam]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Harry Putnam <reader@newsguy.com> writes:
> 
> > OOPs ... F does do something usefull when unplugged, but what does "g"
> > do when unplugged.
> 
> The same as when it's plugged -- it reads the active file and stuff.
> Of course, if you only update the Agent active file from the same Gnus 
> that you're reading unplugged, there won't be anything new in there,
> but I use a separate Gnus to fetch news/mail from the one I read in.
> The one I read news in is always unplugged.

I sense that I'm missing out on some nice functionality. How does the
updated active file, new messages etc get to the unplugged
version. Probably something obvious but I've experimented a few times
with 2 running instances of gnus, also with M-x gnus-slave but must
have missed out on the right technique.

Might be a good idea if some of you old (young) hands spelled out some
of your cool tricks for us new (old) hands when time permits.

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Harry Putnam <reader@newsguy.com> writes:

> I sense that I'm missing out on some nice functionality. How does the
> updated active file, new messages etc get to the unplugged
> version.

Er, well, you press `g'.  :-)

Ok, here's what I do.

In the Emacs that I'm reading news in, I start with:

(defun pgnus-unplugged ()
  (interactive)
  (setq gnus-plugged nil)
  (gnus))

And then I just read, and answer, and stuff.  I never plug in.  Well,
I do sometimes, but usually not.

In the other Emacs, I call the following command:

(defun lars-fetch-news ()
  (interactive)
  (push "~/pgnus/lisp" load-path)
  (let ((init-file-user "")
	(gnus-always-read-dribble-file t))
    (gnus))
  (gnus-group-send-drafts)
  (gnus-agent-fetch-session)
  (gnus-group-quit))

I call this command whenever I want to send/get new mail/news.

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

Re: Song of Programming

[Neil Crellin]

Harry Putnam <reader@newsguy.com> writes:
> I sense that I'm missing out on some nice functionality. How does the
> updated active file, new messages etc get to the unplugged
> version. Probably something obvious but I've experimented a few times
> with 2 running instances of gnus, also with M-x gnus-slave but must
> have missed out on the right technique.

Somewhat tangential, but you reminded me of a wishlist item I'd been
meaning to raise.  How is one supposed to be able to run a gnus-slave
unplugged safely?  A gnus-slave-unplugged function which ensured that
both slave and gnus-unplugged were set correctly would be nice here.

Re: Song of Programming

[Jan Vroonhof]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Before that, I'll be answering mail and stuff, but are there any
> particular outstanding issues that should be resolved before going
> beta and freezing features?  (Except fixing the usual bugs?)

I have not installed the latest pgnus'en, so I'm not sure whether you
already. But I would love it if you could go back to using
insert-file-contents-literally for loading attachments into
buffers. Yes I know current versions will not work with ange-ftp/efs,
but I think I think it is a small price to pay for correctness.
Let the (X)Emacs maintainers fix the remote file problem.

Jan

Re: Song of Programming

[Harry Putnam]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Harry Putnam <reader@newsguy.com> writes:
> 
> > I sense that I'm missing out on some nice functionality. How does the
> > updated active file, new messages etc get to the unplugged
> > version.
> 
> Er, well, you press `g'.  :-)

No wonder I missed it!  Its one of those ultra esoteric, two handed
stealth commands that emacs is famous for.   : )

> Ok, here's what I do.
> 
> In the Emacs that I'm reading news in, I start with:
> 
> (defun pgnus-unplugged ()
>   (interactive)
>   (setq gnus-plugged nil)
>   (gnus))
> 
> And then I just read, and answer, and stuff.  I never plug in.  Well,
> I do sometimes, but usually not.
> 
> In the other Emacs, I call the following command:
> 
> (defun lars-fetch-news ()
>   (interactive)
>   (push "~/pgnus/lisp" load-path)
>   (let ((init-file-user "")
> 	(gnus-always-read-dribble-file t))
>     (gnus))
>   (gnus-group-send-drafts)
>   (gnus-agent-fetch-session)
>   (gnus-group-quit))
> 
> I call this command whenever I want to send/get new mail/news.

Thanks... this looks nifty

Re: Song of Programming

[Robert Epprecht]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Harry Putnam <reader@newsguy.com> writes:
>
> > The fact that gnus-agent looses track of agentized posts once they are
> > expunged from the server is a singularly annoying aspect of gnus, for
> > us offliners.
>
> Yup.  Fix in Pterodactyl Gnus v0.91.

Wow! Thank you Lars, that's a bit like christmas for me...

So now there's only one thing left, to make the agent a *very*
handy offline tool: a better interface to the download score rules.

It would be very convenient if you could say 'I like this thread,
author or whatever, please download all the articles' Having to
edit a score file by hand or marking each single posting is a bit
too complicated. On the other hand we have this very good interface
to the other score files. So why not use the normal scoring rules
for downloading? This very likely wouldn't be too much work to
implement. Or have gnus-agent-use-normal-scoring-for-download
or something like that to swich it on or off?
Of course it wouldn't make sense to score on the body, so maybe
the agent should ignore this or warn, if someone tries to do so?

Another solution would be to do something like having the scoring
interface understand the 'symbolic prefix', and let there be values
like 'a' for all-score-rules or 'd' for download-score-rules.
So you could say something like 'M-i d I a s p' or so to change the
agents download score rules.

Anyway, a handy interface to the download score rules would be a very
substantial improvement of the agents abilities.

Robert Epprecht

Re: Song of Programming

[Harry Putnam]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> In the other Emacs, I call the following command:
> 
> (defun lars-fetch-news ()
>   (interactive)
>   (push "~/pgnus/lisp" load-path)
>   (let ((init-file-user "")
> 	(gnus-always-read-dribble-file t))
>     (gnus))
>   (gnus-group-send-drafts)
>   (gnus-agent-fetch-session)
>   (gnus-group-quit))

Should I be getting a message like:

Quit reading news without saving .newsrc? [yes no]

When ever I call your function above?

Re: Song of Programming

[Kai.Grossjohann]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Before that, I'll be answering mail and stuff, but are there any
> particular outstanding issues that should be resolved before going
> beta and freezing features?  (Except fixing the usual bugs?)

Small issue: write functions for the MML / Insert menu.  I think the
rest of the MML menu already works, right?

I think having something for editing MML tags is not so important, yet
would be nice to have.

kai
-- 
Life is hard and then you die.

Re: Song of Programming

[Alexandre Oliva]

On Jul  3, 1999, Per Abrahamsen <abraham@dina.kvl.dk> wrote:

> Alexandre Oliva <oliva@dcc.unicamp.br> writes:
>> I'd really appreciate to have the fancy splitting patch I posted
>> installed in the next release, as well as the group-params based mail
>> splitting mechanism.  

> I'd like those kind of changes to have a longer alpha testing period,
> Errors in the mail splitting code can easily lead to lost mail, even
> in ways that aren't discovered for a while.

When there's an error in mail splitting, messages go to a `bogus', so
mail is not lost (as I've learned while implementing these changes).
Moreover, the changes are simple enough that I don't see much risk in
adopting them.  Anyway, I'm not the one to decide...  I'd be glad to
have *some* gnus version supporting this feature, even if not 5.8.0,
so that I don't have to keep maintaining a modified file locally.

But I've received enough positive feedback after posting this patch
that I think it would be worth having it in the next stable release.
This also indicates that I'm not the only one using it (but not that
it's been widely tested).  The only way to have it tested widely is to 
install it in some (alpha) release of gnus :-)

-- 
Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il
{oliva,Alexandre.Oliva}@dcc.unicamp.br  aoliva@{acm.org,computer.org}
oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org}
*** E-mail about software projects will be forwarded to mailing lists

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Neil Crellin <neilc@wallaby.cc> writes:

> Somewhat tangential, but you reminded me of a wishlist item I'd been
> meaning to raise.  How is one supposed to be able to run a gnus-slave
> unplugged safely?  A gnus-slave-unplugged function which ensured that
> both slave and gnus-unplugged were set correctly would be nice here.

Hm.   Is that something that one would do much?

I'm a bit wary of adding commands that would multiply exponentially as 
features are added...

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

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Jan Vroonhof <vroonhof@math.ethz.ch> writes:

> I have not installed the latest pgnus'en, so I'm not sure whether you
> already. But I would love it if you could go back to using
> insert-file-contents-literally for loading attachments into
> buffers.

This has been fixed, I think.

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

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Robert Epprecht <epprecht@sunweb.ch> writes:

> So why not use the normal scoring rules for downloading? This very
> likely wouldn't be too much work to implement.

I think it might be much work -- the current scoring code depends on
certain things being set up to do its thing, and those things aren't
set up unless you're entering a group.  (Stuff like threading and the
like.)

I haven't looked closely at it, though.

> Another solution would be to do something like having the scoring
> interface understand the 'symbolic prefix', and let there be values
> like 'a' for all-score-rules or 'd' for download-score-rules.
> So you could say something like 'M-i d I a s p' or so to change the
> agents download score rules.

That sounds like a good idea...

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

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Harry Putnam <reader@newsguy.com> writes:

> Quit reading news without saving .newsrc? [yes no]
> 
> When ever I call your function above?

Yes, but I have `gnus-expert-user' set, so I never get those queries.

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

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Kai.Grossjohann@CS.Uni-Dortmund.DE writes:

> Small issue: write functions for the MML / Insert menu.  I think the
> rest of the MML menu already works, right?

The two commands in the MML / Insert menu seem to work, but there
should probably be more of them?

> I think having something for editing MML tags is not so important, yet
> would be nice to have.

It would...

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

Re: Song of Programming

[Neil Crellin]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
> Neil Crellin <neilc@wallaby.cc> writes:
> 
> > Somewhat tangential, but you reminded me of a wishlist item I'd been
> > meaning to raise.  How is one supposed to be able to run a gnus-slave
> > unplugged safely?  A gnus-slave-unplugged function which ensured that
> > both slave and gnus-unplugged were set correctly would be nice here.
> 
> Hm.   Is that something that one would do much?
>
> I'm a bit wary of adding commands that would multiply exponentially as 
> features are added...

I, for one, would use that a lot.  To give you an idea of the sort of
usage scenario I'm thinking of: I frequently have certain mail groups
with large backlogs I need to slog through, but want to be able to
continue to read new mail that comes in in a timely fashion, or I just
want to be able to take a break from the backlog slog and browse some
of my other groups without incurring the huge penalty of exiting and
re-entering the groups, expiring and generating new sorted scored
summary buffers for the big backlog group.  gnus-slave is ideal for
this sort of purpose, but only when plugged.  I can have two emacs
windows with one dedicated to the monster backlog, and the other for
taking breaks and catching up on other stuff without pain in the big
group window. 

To get the same sort of functionality unplugged is a challenge.
Currently, I set up a gnus master and a gnus-slave plugged, then J j
both of them to go unplugged and drop the connection.  Then I can
carry the laptop around the house or far away from the net connection.
I'd like to be able to do that without the convolution.

Currently gnus-plugged and gnus-unplugged just call (gnus) bare as
their last sub-call.  gnus-slave on the other hand accepts (& optional
arg) and invoke gnus with eg (gnus arg nil 'slave).  So one suggestion
is should gnus-unplugged and gnus-plugged take (&optional arg) and do
similar?  

Function gnus already handles dont-connect and slave as arguments. 
Since gnus-plugged and gnus-unplugged are just calls to gnus with
gnus-plugged set to t or nil beforehand, would it even make sense to
absorb gnus-plugged as an argument like those to gnus itself?  Or does
that only make sense for folks who've agentized?

Will this do what I need in the meantime?

(defun gnus-unplugged-slave (&optional arg)
  "Start Gnus slave unplugged."
  (interactive)
  (setq gnus-plugged nil)
  (gnus arg nil 'slave))

-- 
Neil Crellin <neilc@wallaby.cc>

Re: Song of Programming

[Didier Verna]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Before that, I'll be answering mail and stuff, but are there any
> particular outstanding issues that should be resolved before going
> beta and freezing features?  (Except fixing the usual bugs?)

        There's another really important issue that 's still unresolved: the
fact that stuff in gnus-xmas.el overrides the user custom settings for some
variables.

-- 
    /     /   _   _       Didier Verna        http://www.inf.enst.fr/~verna/
 - / / - / / /_/ /     ENST, INFRES C201.1       mailto:verna@inf.enst.fr
/_/ / /_/ / /__ /        46 rue Barrault        Tel.   +33 (1) 45 81 73 46
                       75013 Paris, France      Fax.   +33 (1) 45 81 31 19

Re: Song of Programming

[Harry Putnam]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Harry Putnam <reader@newsguy.com> writes:
> 
> > Quit reading news without saving .newsrc? [yes no]
> > 
> > When ever I call your function above?
> 
> Yes, but I have `gnus-expert-user' set, so I never get those queries.

Trying to be like the big boys, I set `gnu-expert-user' to t. Now when
calling  lars-fetch-news  the fetching, sending etc all happens then
gnus silently shuts clear down.  Writing .newsrc.eld as it goes.

Err . . not what I had hoped for.

Re: Song of Programming

[Kai.Grossjohann]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Before that, I'll be answering mail and stuff, but are there any
> particular outstanding issues that should be resolved before going
> beta and freezing features?  (Except fixing the usual bugs?)

IMVHO the interface for `v' on a MIME button is rather awful :-|
Excuse me if I use such strong words.

OT1H it pretends to have completion (SPC tries to complete), OTOH it
requires you to enter stuff like "xv %s" for viewing.

The easy way to fix this is to turn off completion for entering the
right commands.  A possibly better way to fix this is to let the user
enter the MIME type and to choose the corresponding viewing command.

Maybe the best is to offer both: one command which says `view this
part normally but pretend it has a different content type' (this would
be useful for mails received from broken mailers which are unable to
supply a meaningful content type), and another command which allows
the user to specify the viewing command completely, even if nothing
like it is in the .mailcap file (this would be useful if one normally
uses ee to view images but wants to use xv for a specific image).  Or
would it be better to offer a command like ! in dired?  (There, you
just enter the command and the file name is tacked on at the end.  If
necessary, you can put * somewhere in the command, which is replaced
by the file name.)

kai
-- 
Life is hard and then you die.

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Neil Crellin <neilc@wallaby.cc> writes:

> Function gnus already handles dont-connect and slave as arguments. 
> Since gnus-plugged and gnus-unplugged are just calls to gnus with
> gnus-plugged set to t or nil beforehand, would it even make sense to
> absorb gnus-plugged as an argument like those to gnus itself?  Or does
> that only make sense for folks who've agentized?

It only makes sense for the agentized people.

> Will this do what I need in the meantime?
> 
> (defun gnus-unplugged-slave (&optional arg)
>   "Start Gnus slave unplugged."
>   (interactive)
>   (setq gnus-plugged nil)
>   (gnus arg nil 'slave))

Yup; I think so.

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

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Didier Verna <verna@inf.enst.fr> writes:

>         There's another really important issue that 's still unresolved: the
> fact that stuff in gnus-xmas.el overrides the user custom settings for some
> variables.

Which variables?

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

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Harry Putnam <reader@newsguy.com> writes:

> Trying to be like the big boys, I set `gnu-expert-user' to t. Now when
> calling  lars-fetch-news  the fetching, sending etc all happens then
> gnus silently shuts clear down.  Writing .newsrc.eld as it goes.
> 
> Err . . not what I had hoped for.

Well...  Why not?

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

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Kai.Grossjohann@CS.Uni-Dortmund.DE writes:

> OT1H it pretends to have completion (SPC tries to complete), OTOH it
> requires you to enter stuff like "xv %s" for viewing.

Yeah.  I think it's primarily meant to allow you to choose between all 
the possible registered viewers for a certain MIME type -- you're not
supposed to write the command explicitly.  (For that you have `|' and
the like.)  So I think it makes sense to have `v' allow
auto-completion. 

> Maybe the best is to offer both: one command which says `view this
> part normally but pretend it has a different content type' (this would
> be useful for mails received from broken mailers which are unable to
> supply a meaningful content type),

A new command for viewing as a different media type sounds nice, but
would that be useful?

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

Re: Song of Programming

[Didier Verna]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Didier Verna <verna@inf.enst.fr> writes:
> 
> >         There's another really important issue that 's still unresolved: the
> > fact that stuff in gnus-xmas.el overrides the user custom settings for some
> > variables.
> 
> Which variables?

        We've talked about this some time ago. The problem is in
`gnus-xmas-redefine'. Called after the custom file has been read,
customizations are gone. I think what you actually want is to `defcustom'
stuff differently according to which flavor of Emacs you're in. At the time we 
talked about this problem, I could see three solutions IIRC:

- either you provide two files with the same defcustom's but with different
values, one for each flavor of Emacs. Only one of these files is loaded then.
This solution has problems though, for example if some of the customized stuff 
has an autoload cookie.

- or you enclose each concerned defcutom in a test on emacs-version (which
actually means that you will have also 2 defcustom's per variable, but in the
same file). However, you're likely to get problems of the same sort (autoloads 
etc).

- or you really have only one defcustom per variable, remove gnus-xmas (at
least the problematic part) and put the xemacs-specifi-stuff calls in the
custom default value, which would be defined but no-op under GNU Emacs.

        I think the third solution is the best.

-- 
    /     /   _   _       Didier Verna        http://www.inf.enst.fr/~verna/
 - / / - / / /_/ /     ENST, INFRES C201.1       mailto:verna@inf.enst.fr
/_/ / /_/ / /__ /        46 rue Barrault        Tel.   +33 (1) 45 81 73 46
                       75013 Paris, France      Fax.   +33 (1) 45 81 31 19

Re: Song of Programming

[Kai.Grossjohann]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Kai.Grossjohann@CS.Uni-Dortmund.DE writes:
> 
> > OT1H it pretends to have completion (SPC tries to complete), OTOH it
> > requires you to enter stuff like "xv %s" for viewing.
> 
> Yeah.  I think it's primarily meant to allow you to choose between all 
> the possible registered viewers for a certain MIME type -- you're not
> supposed to write the command explicitly.  (For that you have `|' and
> the like.)  So I think it makes sense to have `v' allow
> auto-completion. 

Oh.  I haven't tried completion for a *long* time now.  Now I see that
it actually offers a meaningful completion list.  Never mind, then.

> > Maybe the best is to offer both: one command which says `view this
> > part normally but pretend it has a different content type' (this would
> > be useful for mails received from broken mailers which are unable to
> > supply a meaningful content type),
> 
> A new command for viewing as a different media type sounds nice, but
> would that be useful?

I think some broken mailers exist which tag everything as
application/octet-stream and require the recipient to tell the file
type from the file name.

kai
-- 
Life is hard and then you die.

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Kai.Grossjohann@CS.Uni-Dortmund.DE writes:

> I think some broken mailers exist which tag everything as
> application/octet-stream and require the recipient to tell the file
> type from the file name.

And there are people who send out text/html without a Content-Type
header.  So there should be a command to view parts as different media
types.  Should it just allow completion over all known types, and then
take the default action?

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

Re: Song of Programming

[Wes Hardaker]

>>>>> On 03 Jul 1999 07:28:23 -0700, Harry Putnam <reader@newsguy.com> said:

Harry> The fact that gnus-agent looses track of agentized posts once
Harry> they are expunged from the server is a singularly annoying
Harry> aspect of gnus, for us offliners.

Similarly, "J s" "g" "[Enter]" causes problems with lost messages,
unless thats been fixed recently (I'm currently in on-line mode all
the time till I can play with nnimap some more to get its off-line
support more dependable).

-- 
"Ninjas aren't dangerous.  They're more afraid of you than you are of them."

Re: Song of Programming

[Kai.Grossjohann]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> And there are people who send out text/html without a Content-Type
> header.  So there should be a command to view parts as different media
> types.  Should it just allow completion over all known types, and then
> take the default action?

That's the easiest for the user, I think, even if not as flexible as
allowing the user to enter an arbitrary command.

kai
-- 
Life is hard and then you die.

Re: Song of Programming

[Harry Putnam]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Harry Putnam <reader@newsguy.com> writes:
> 
> > Trying to be like the big boys, I set `gnus-expert-user' to t. Now when
> > calling  lars-fetch-news  the fetching, sending etc all happens then
> > gnus silently shuts clear down.  Writing .newsrc.eld as it goes.
> > 
> > Err . . not what I had hoped for.
> 
> Well...  Why not?

Ok... so you're saying you *want* gnus to shutdown each time you call
lars-fetch-news ... press M-x gnus thereafter to restart if you want
to fetch again later.  All news and mail that was fetched is then
called up in the unplugged instance by pressing 'g'.

I get the mail that was incorporated by last `lars-fetch-news' but
never see the new news that was downloaded.

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Didier Verna <verna@inf.enst.fr> writes:

>         We've talked about this some time ago. The problem is in
> `gnus-xmas-redefine'.

That function only adds functions to a bunch of hooks and redefines
some other functions.  It doesn't alter any (normal) variables...

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

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Wes Hardaker <wjhardaker@ucdavis.edu> writes:

> Similarly, "J s" "g" "[Enter]" causes problems with lost messages,
> unless thats been fixed recently (I'm currently in on-line mode all
> the time till I can play with nnimap some more to get its off-line
> support more dependable).

I've never had a problem with that key sequence.  Under what
circumstances does it fail?

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

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Harry Putnam <reader@newsguy.com> writes:

> Ok... so you're saying you *want* gnus to shutdown each time you call
> lars-fetch-news ...

Yes.

> press M-x gnus thereafter to restart if you want
> to fetch again later.

No, I never `M-x gnus' in the Emacs that I fetch news with.

> I get the mail that was incorporated by last `lars-fetch-news' but
> never see the new news that was downloaded.

What is the value of your `gnus-read-active-file' variable?

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

Re: Song of Programming

[Didier Verna]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Didier Verna <verna@inf.enst.fr> writes:
> 
> >         We've talked about this some time ago. The problem is in
> > `gnus-xmas-redefine'.
> 
> That function only adds functions to a bunch of hooks and redefines
> some other functions.  It doesn't alter any (normal) variables...

        Yes it does, because by the time gnus-xmas-redefine is called,
gnus-group (and probably others) has not been loaded, so for example
gnus-group-mode-hook has not been defcustom'ed and doesn't have any value as a
variable. Calling add-hook on it thus *sets* it to the value and your
customization is gone.

        The brute force solution is to put a (require 'gnus-group) somewhere
at the beginning of gnus-xmas. This must probably be done also for other libs
not yet loaded.

-- 
    /     /   _   _       Didier Verna        http://www.inf.enst.fr/~verna/
 - / / - / / /_/ /     ENST, INFRES C201.1       mailto:verna@inf.enst.fr
/_/ / /_/ / /__ /        46 rue Barrault        Tel.   +33 (1) 45 81 73 46
                       75013 Paris, France      Fax.   +33 (1) 45 81 31 19

Re: Song of Programming

[Wes Hardaker]

>>>>> On 08 Jul 1999 09:00:35 +0200, Lars Magne Ingebrigtsen <larsi@gnus.org> said:

Lars> Wes Hardaker <wjhardaker@ucdavis.edu> writes:
>> Similarly, "J s" "g" "[Enter]" causes problems with lost messages,
>> unless thats been fixed recently (I'm currently in on-line mode all
>> the time till I can play with nnimap some more to get its off-line
>> support more dependable).

Lars> I've never had a problem with that key sequence.  Under what
Lars> circumstances does it fail?

It used to (again, I haven't tested it in a long time):

1) J s would get all the articles, say from 1-20.
2) g would update the group so that 21-25 were known.
3) Enter would then enter the group, showing the 1-20 articles.
4) Exiting the group with catchup would then catchup 1-25, which is
   erroneous since I never saw them.

-- 
"Ninjas aren't dangerous.  They're more afraid of you than you are of them."

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Kai.Grossjohann@CS.Uni-Dortmund.DE writes:

> That's the easiest for the user, I think, even if not as flexible as
> allowing the user to enter an arbitrary command.

I've now added this.

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

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Wes Hardaker <wjhardaker@ucdavis.edu> writes:

> 1) J s would get all the articles, say from 1-20.
> 2) g would update the group so that 21-25 were known.
> 3) Enter would then enter the group, showing the 1-20 articles.
> 4) Exiting the group with catchup would then catchup 1-25, which is
>    erroneous since I never saw them.

That hasn't been fixed, so the moral is -- don't `g', then go offline
and read articles.  If you `g' online, you must `J s' afterwards.

It should be fixed, but it's not a high priority.

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

Re: Song of Programming

[Matt Pharr]

I'm not sure if anyone has brought this up yet, but what about bundling the 
wonderfully useful nnir?  The main issue, IMHO, is that only nnml and nnmh
are well-supported--must it be totally general to be included in the
distribution?

-matt
-- 
Matt Pharr                                   mmp@graphics.stanford.edu
<URL:http://graphics.stanford.edu/~mmp>

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Matt Pharr <mmp@Graphics.Stanford.EDU> writes:

> I'm not sure if anyone has brought this up yet, but what about bundling the 
> wonderfully useful nnir?  The main issue, IMHO, is that only nnml and nnmh
> are well-supported--must it be totally general to be included in the
> distribution?

It would be neat if nnir was a more general thing...

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

Re: Song of Programming

[Kai Großjohann]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> It would be neat if nnir was a more general thing...

I completely agree.  Well, right now it supports nnimap and nnmh, nnml
and nnspool.  The problem with supporting other backends lies in
figuring out the article number from the search result.  Hm.

Adding support for other backends with freeWAIS-sf is on my todo
list.  I think Gnus stores the group name (if necessary) and the
article number in extra headers for the other backends, doesn't it?
If this is true, then it would be reasonably simple to tell
freeWAIS-sf to output the article number (and group name) in the
retrieval result.

But I'm not at all sure how to go about this with Glimpse.  Is it
possible to search an nnfolder file (say) with Glimpse and to be able
to tell from the search result the article number?  Seems to be
difficult.  OTOH, Glimpse is quite limited anyway, so...

Would you like to include nnir.el as-is right now, or would you like
to wait until it contains more features (I'm not promising anything
time-wise)?

kai
-- 
Life is hard and then you die.

Re: Song of Programming

[Wes Hardaker]

>>>>> On 09 Jul 1999 19:44:51 +0200, Lars Magne Ingebrigtsen <larsi@gnus.org> said:

Lars> That hasn't been fixed, so the moral is -- don't `g', then go
Lars> offline and read articles.  If you `g' online, you must `J s'
Lars> afterwards.

Lars> It should be fixed, but it's not a high priority.

If its not fixed, it should be pointed out in the manual with extremely 
large bold font size=2000 letters.  I'm happy I noticed it when I was
operating on a newsgroup and not on a mail folder.

I actually consider it a pretty serious bug.  You'll miss mail/news if 
you don't know about it, which is a very bad thing.  And you won't
notice it if the group is large (I noticed it because I was expecting
like 10 articles and only got 5 upon entering the group).

The correct solution would be to not catch up articles that aren't
shown in the summary buffer (ie, if the # unread was 20 and the agent
had only information on 15 of them, then after catchup and exit of
those 15 the # unread should drop to 5 not 0).

-- 
"Ninjas aren't dangerous.  They're more afraid of you than you are of them."

Re: Song of Programming

[Paul Stodghill]

>>>>> On 09 Jul 1999 19:44:51 +0200, Lars Magne Ingebrigtsen <larsi@gnus.org> said:
    Lars> That hasn't been fixed, so the moral is -- don't `g', then go
    Lars> offline and read articles. If you `g' online, you must `J s'
    Lars> afterwards.

    Lars> It should be fixed, but it's not a high priority.

>>>>> "Wes" == Wes Hardaker <wjhardaker@ucdavis.edu> writes:

    Wes> [...] I actually consider it a pretty serious bug. [...]

So do I. Can I safely use this as a workaround?

    (defun pvs-fetch-session-after-new-news ()
      (interactive)
      (when gnus-plugged
        (gnus-agent-fetch-session)))

    (add-hook 'gnus-after-getting-new-news-hook
              'pvs-fetch-session-after-new-news)


-- 
Paul Stodghill <stodghil@cs.cornell.edu>
Dept. of Computer Science, Upson Hall, Ithaca, NY  14853
Phone: 607-254-8838   FAX: 607-255-4428
http://www.cs.cornell.edu/stodghil/

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) writes:

> Would you like to include nnir.el as-is right now, or would you like
> to wait until it contains more features (I'm not promising anything
> time-wise)?

If you think that nnir should be included, I'm all for it.  :-)

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

Re: Song of Programming

[Lars Magne Ingebrigtsen]

Wes Hardaker <wjhardaker@ucdavis.edu> writes:

> If its not fixed, it should be pointed out in the manual with extremely 
> large bold font size=2000 letters.  I'm happy I noticed it when I was
> operating on a newsgroup and not on a mail folder.

I don't think it can happen on mail groups.  Well, it can happen on
nnimap groups, but not on the ones that store mail locally.

> The correct solution would be to not catch up articles that aren't
> shown in the summary buffer (ie, if the # unread was 20 and the agent
> had only information on 15 of them, then after catchup and exit of
> those 15 the # unread should drop to 5 not 0).

Yes.  But if the articles really were cancelled?  How is Gnus to know
the difference?

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

Re: Song of Programming

[François Pinard]

Lars Magne Ingebrigtsen <larsi@gnus.org> écrit:

> Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) writes:

> > Would you like to include nnir.el as-is right now, or would you like
> > to wait until it contains more features (I'm not promising anything
> > time-wise)?

> If you think that nnir should be included, I'm all for it.  :-)

I do not use `nnir' very often, but yet, regularly, and am rather happy
with it.  Sometimes, I have to struggle a bit against errors, but luckily
enough, usually find my way through, sometimes just restarting Emacs.

This is vague, I know, but I would like stability over features, if I had some
right to choose.

The only feature that I would really be happy to have, but it goes somewhat
beyond Gnus, would be the ability to use `nnir' to access both my `nnml'
and external Babyl files, all at once, and be able to revisit my Babyl
files rewritably, just as `nnml' files may be revisited.

-- 
François Pinard   http://www.iro.umontreal.ca/~pinard

Re: Song of Programming

[Wes Hardaker]

>>>>> On 27 Aug 1999 19:16:58 +0200, Lars Magne Ingebrigtsen <larsi@gnus.org> said:

Lars> Wes Hardaker <wjhardaker@ucdavis.edu> writes:
>> If its not fixed, it should be pointed out in the manual with extremely 
>> large bold font size=2000 letters.  I'm happy I noticed it when I was
>> operating on a newsgroup and not on a mail folder.

Lars> I don't think it can happen on mail groups.  Well, it can happen on
Lars> nnimap groups, but not on the ones that store mail locally.

Which is where I read mail.

>> The correct solution would be to not catch up articles that aren't
>> shown in the summary buffer (ie, if the # unread was 20 and the agent
>> had only information on 15 of them, then after catchup and exit of
>> those 15 the # unread should drop to 5 not 0).

Lars> Yes.  But if the articles really were cancelled?  How is Gnus to know
Lars> the difference?

Hmm.   Good point.

(Of course I *think* I know what we were talking about here...  The
original thread has fallen beyond my expirory time)

-- 
"Ninjas aren't dangerous.  They're more afraid of you than you are of them."