Re: Back to Gnus?

[Dave Abrahams <dave@boostpro.com>]

At Tue, 05 Apr 2011 18:49:33 -0400,
Vitaly Mayatskikh wrote:
> 
>  Vitaly Mayatskikh <v.mayatskih@gmail.com>
> Subject: Re: Back to Gnus?
> Date: Tue, 05 Apr 2011 18:49:33 -0400
> To: wl-en@lists.airs.net
> Reply-To: wl-en@lists.airs.net
> Message-Id: <87ei5guxdu.wl@gravicappa.localhost.localdomain>
> User-Agent: Wanderlust/2.15.9 (Almost Unreal) Emacs/23.3 Mule/6.0
>  (HANACHIRUSATO)
> List-Post: <mailto:wl-en@lists.airs.net>
> 
> At Tue, 05 Apr 2011 18:29:50 -0400, Dave Abrahams wrote:
> 
> > > >I'm not happy about it, but I'm having enough problems with WL that it
> > > >prompts serious consideration of a return to Gnus.  Among the issues:
> > > >
> > > >* Very rough handling of mail composition
> > > >
> > > >  * New messages with citations are created with embedded mime buttons
> > > 
> > > You can press 'C-c m' (wl-summary-toggle-mime-buttons) before you
> > > reply to a message.
> > 
> > Yes, I know, but what a pain! 
> 
> (setq mime-view-buttons-visible nil)

And how does that help when I need to see MIME parts?

Just getting WL to display MIME reasonably has been an adventure, and
I'm *still* encountering quirks I never had with Gnus.  It doesn't
"just work" out-of-the-box; not even close.

> > Especially given my last issue of hangs, this becomes just one of
> > many little inconveniences that add up to cost me time and
> > aggravation every day.
> > 
> > Oh, and here's another thing: those keybindings don't work when you're
> > actually in the message highlighting the portion you want cited.  They
> > only work in the summary, and that causes aggravation.  Similarly,
> > there are commands/keybindings that work in the message itself that
> > should also work from summary mode (e.g. decrypt PGP section).
> 
> You have Wanderlust sources, right? :)

Yes, of course.  And I've poured lots of my time into hacking around
in there over the past year or two.

> > > >  * Font lock doesn't work reliably, at all.  In fact, it's incredibly
> > > >    annoying.  For example, If I want to insert a line a the beginning
> > > >    of a message with a citation, I go to the end of the line
> > > >    containing "--text follows this line--", hit `RET', and the
> > > >    highlighting from that line extends to cover everything I type.
> > > >    The same goes when inserting in cited text.
> > > 
> > > Yes, this is annoying, but you can turn off highlighting of
> > > mail-header-separator in wl/wl-highlight.el. Just remove the lines
> > > 179-188:
> > > 
> > > (wl-defface wl-highlight-header-separator-face
> > >   '(
> > >   (((type tty)
> > > [...]
> > 
> > Right; I have to modify WL.  And I like the mail-header-separator to
> > be highlighted.  I could live with giving up mail-header-separator
> > highlighting, but that doesn't help when I am inserting text into a
> > cited section; I still get the highlighting of the citation on my new
> > text.  The latter is by far the more common of these cases for me.
> 
> This issue was worked around in this mail list, btw:
> 
> (defun my-highlight-hook (beg end len)
>   (let ((begining (or (save-excursion
> 			(goto-char beg)
> 			(re-search-backward "^" nil t))
> 		      (point-min)))
> 	(ending (or (save-excursion
> 		      (goto-char end)
> 		      (re-search-forward "$" nil t))
> 		    (point-max))))
>     (put-text-property begining ending 'face nil)
>     (wl-highlight-message begining ending t nil)
>     (wl-highlight-message begining ending t t)))
> 
> (add-hook 'wl-draft-mode-hook
>           (lambda ()
>             (make-local-variable
>              'after-change-functions)
>             (add-hook 'after-change-functions
>                       'my-highlight-hook)))
> 

Right, almost works perfectly (still leaves the background color on
the end of the line I'm editing) ... provided I can afford to
re-highlight the entire buffer after every keypress.  If this is what
you're expected to do to make it work, why isn't that built in?

> > > >* Generally, too much customization required just to get some
> > > >  functionality that's de-rigeur in many mail clients (e.g. signature
> > > >  support).  Just look at the size of my customizations!
> > > >  https://github.com/dabrahams/dwamacs/blob/master/settings/wl-settings.el
> > > >
> > > >* Connections to my IMAP and SMTP server drop every few minutes and WL
> > > >  hangs instead of dealing with it gracefully.  I need to `C-g', wait
> > > >  for Emacs to notice, then `M-t M-t' to get back to normal.
> > > 
> > > Hmm, I use WL with three different IMAP servers (Gmail, Fastmail and
> > > Exchange 2010) and I don't have this problem.
> > 
> > I am also connecting to the GMail IMAP server.  I'm not sure why I am
> > losing the connection (actually, I'm just assuming that's the root
> > cause of this problem), but the fact is that WL doesn't deal with it
> > gracefully.  And that problem also occurs every time I disconnect my
> > wired connection and take my laptop to a different area or wake the
> > laptop from sleep.  I'm sure that happens to you, too.
> 
> That can be solved easily, if somebody really wants to solve it.

You're saying all of those related problems can be solved easily?

If it were really going to take *me* less than a couple hours, that,
combined with your almost-functionally-perfect-but-inefficient fix for
highlighting would probably make it worth sticking with WL.  But I
fear it would take one whole day of hacking---minimum, including
re-learning how to Grok luna---for me to get that fixed.

> You can't solve the fact that Gnus is only good for NNTP :) So, its
> IMAP sucks, 

My understanding is that in Emacs24 its IMAP is pretty darned good.
I'm going to try it and see.

> offline support sucks, 

I didn't find offline support to suck any worse than it does for WL.
What's the problem?

> sources organization sucks, code
> quality sucks,... sucks, ... sucks, ... sucks, ... ;)

+1

But if it works and I don't have to hack it, that part doesn't matter.
There are enough other people who care to keep it more-or-less
functional.

As for hackability, WL uses a non-standard undocumented OOP system and
has almost no internal comments and no docstrings.  Working on WL has
been no picnic for me.  As for functionality, WL is painfully slow for
very large groups, of which I have quite a few (both mail and NNTP).

> I've fixed everything worried me in WL without feeling pain in my ass,
> and I'm pretty happy about it. It made ever greater difference,
> because I came to WL from Gnus, and I will never return back. No
> way. Gnus sucks, it has never worked well enough for me. 

That's exactly how I felt after arriving at WL from Gnus.  But
combined with rumored nnimap improvements and the problem of hanging
on dropped connections, the other inconveniences are starting to tip
the scales back for me.  This is a somewhat desperate plea to help me
avoid investing in Gnus.

> I want working IMAP and clean, readable sources to customize things
> how I'd like.

I appreciate that, really.  I wish I was as good at working on
undocumented elisp as you are.

-- 
Dave Abrahams
BoostPro Computing
http://www.boostpro.com