Re: [Fwd: [evolution-patches] [resend] patches for #20672 (gtkhtml and mailer)]

[Radek Doulík <rodo ximian com>]

On Thu, 2003-11-13 at 13:33, Not Zed wrote:
> Been a bit busy to check this list closely (didn't realise it was
> mailer related at first).
> 
> Anyway, my 5c worth ...
> 
> On Wed, 2003-11-12 at 06:05, Michel Dänzer wrote: 
> > On Tue, 2003-11-11 at 22:28, Radek Doulík wrote:
> > > 
> > > looks like you aim at two flies at once. If I understand it well, there
> > > are two things:
> > > 
> > >      1. stop composer autosaving when there are no changes from last
> > >         save
> > > 
> > >      2. make evolution restore all composers after crash/shutdown (even
> > >         if they don't contain any changes from initial and/or saved
> > >         state)
> > > 
> > > 1st is addressed by my patch and fixes an actual bug. That I would like
> > > to be reviewed by Jeff/Michael (Cc-ing you as it looks like you missed
> > > that thread?) and go to 1.4.x.
> 
> I'm not sure we care about this enough in 1.4.x, to be honest.  Do
> we?  1.4.x is good enough in this respect.  Auto-saving is fast enough
> not to notice it.

okie, I will apply it only to HEAD

> > > 2nd is rather new feature, so it should be discussed and eventually go
> > > to 2.0.
> > 
> > Why is that a new feature? When I open a composer, and then evo crashes,
> > why shouldn't the composer be restored the next time, even if I didn't
> > make any changes in it? That's a bug in my book.
> Well, FWIW I would find it pretty weird to behave like this.  If you
> haven't actually written anything, then you haven't lost anything
> either.  Does your evolution really crash that often?  Mine certainly
> doesn't.
> 
> Although on the other hand, i have no objection to this changed
> behaviour.

I see one scenario where that can be useful: you read a lot of mail and
don't want to reply immediately. so you just open reply composer and
continue reading mail. you will end up with a lot of open composers
after you finish reading. then you start subsequently writing replies.
if it crashes (which is unlikely ;) you just restart evo. also when you
have to interrupt it, you may start evo later and continue replying.

Is the above scenario you have on mind Michel? Or do you have another
one too?

> > > I think the point 2. of your test case should autosave immediately
> > > before saving to get consistent behavior. (if evo crashes between save
> > > draft and autosave one minute later, composer will not be recovered on
> > > evo start) 
> > 
> > Well, actually it will, but the version autosaved shortly before saving.
> Which is a bug, by any measure.  It should always be the last version
> saved.  Givne you're hitting i/o anyway its an ideal time to schedule
> an auto-save too.
> 
> > > That new feature makes sense to me. As you pointed out it will need also 
> > > changes to evolution closing code - to be able quit without closing all 
> > > the composer windows. (should fire question dialog asking whether you 
> > > want to quit evolution with unfinished composers and recover them on 
> > > next start)
> > 
> > Actually, I'd like it to behave again as 1.2 (or was it way back in 1.0?
> > :) used to, i.e. keep composers autosaved on exit and recover them on
> > startup, no questions asked. A kind of session management for composers,
> > if you will.
> No you don't, 1.2 (or maybe it was 1.0) didn't recover ANY saved
> sessions till you started to write a new mail!
> 
> I'd be pretty surprised if anything changed along the way other than
> improvements anyway.
> 
> > Anyway, I can cope as long as you don't take --force-shutdown away from
> > me as well. ;)
> Don't worry, that'll be around, although i just use ctrl-c myself.
> 
> FWIW I like Radek's patch better, since it is a little simpler.  Maybe
> the missing features from it can be added to finish it off?

sounds good. Not Zed, can you discuss the point 2. on today's meeting?
(I will also mention it to status report I am writing for JP as I will
be missing on the meeting)

Cheers
Radek