Jean-Marc Lasgouttes <[EMAIL PROTECTED]> writes:
>>>>>> "Lars" == Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes:
>
| Lars> I did not know that it would break...
>
| I can believe that.
I am also a bit sad that it did... because I really like the boost
version a lot more than my own.
| So you are leading a project which is still using xforms as interface
| (and not too much annoyed by it as far as I can tell) and you say _we_
| are conservative? It is just that we do not have the same priorities.
Ooh... I am annoyed with xforms... I'd be glad to get rid of it 1.0 or
no 1.0. How long were we promissed a open source xforms? How many
times has the 1.0 released been promissed now?
I just do not trust that project...
OTOH, to switch we need something better to switch to, and the
framework for making the switch and cleanup possible. We are beginning
to get closer to that now, and I rejoice at the mere thought.
| Lars> the issue with lyxstring will just come up more and more often.
| Lars> It is a hack and far from standard.
>
| I can understand that.
That same issue will come up more and more with exceptions as well,
besides I turned exceptions on some days ago and have not received any
bad flak yet... (how come?)
| Lars> ... my lyx compiles so slowly with <string>... I demand
| Lars> lyxstring... Bah! Help make lyx compile faster with standard
| Lars> libraries instead. We have a lot of badly designed code that
| Lars> makes compilation slower.
>
| You've been saying that for ages, but one of the main consquence of
| aggressive templatification of lyx for now has been that it grows both
| in disk space and developpers time. I you shave both by a factor of 2,
| for example, I'll be glad to drop lyxstring :)
I saw some report at the gcc mailing list today that binaries using
lots of templates got reduced by a factor of 5 during the last 24h.
Not sure what this is about... and it will not come until 3.2. anyway.
Anyway... please also remember _why_ lystring (and LString) was
introduced: to that we could use something the was resembling a real
standard string (and something that was not "char const *"). Now
almost all compilers have <string> that surpasses us when it comes to
being standard.
| Seriously, the time and space of 1.2.0 is much higher than 1.1.6,
| which itself had already a lot of stl in it (yes, I know I do not give
| numbers). So the redesign you are alluding too should be rather
| drastic to be effective. I see how you can win 100k and a couple minutes
| here and there,
If you see that, please tell me pronto, it is not like my compiles are
fast either...
| but it won't be enough to offset the bloat drift we
| will gain in the same time. And most of my lyx work will be done on a
| pII 366 laptop.
Yes, and I understand that this is cumbersome...,
(and especially on a laptop where the io-subsystem usually has a lot
to be wanted)
--
Lgb
