Subject: Re: bool accepted == true ? (was Re: Abi string class)
From: Thomas Fletcher (thomasf@qnx.com)
Date: Thu Feb 01 2001 - 11:00:14 CST
On Thu, 1 Feb 2001, Mike Nordell wrote:
> Joaquin Cuenca Abela wrote:
> > of course it will be cool to use map, vector, the algos, etc. but I
> > think it's better to wait for our "supported" compilers to implement
> > correctly the STL (btw, are there any "supported" compiler (not an
> > imaginary one) that right now fails to implement this stuff?
>
> Yes. MSVC. PJP wasn't able to implement a conforming implementations since
> M$ (the largest software company on earth?) couldn't produce a conforming
> compiler. His implementation was buggy. Btw, it's confirmed they're not
> going to deliver a conforming C++ compiler even with the 7.0. :-<
> There are available bugfixes for the header files at dinkumware site, but M$
> still ships the old buggy stuff.
QNX is going to be using the latest from dinkumware.
> But there is another problem we face. What "supported" compilers do we have?
> Sam, feel free to pitch in any time. We maintain a "least common
> denominator" approach atm which I _do_ approve of, no matter what my
> postings might suggest. But I think we at least once in a while (6 months,
> one year?) should evaluate our position re. these issues.
I think that re-evaluating our position is fine, but I don't think that
this switch to deciding that all of a sudden wham-o namespaces, templates
and any or all of the other "advanced" (all in perspective) features
are required or are needed. I believe that you should look at using
a tool when that tool is required. We have done very well in terms
of a size/speed/programmer burden trade-off in implementing our own
components. I don't really want to have us only use half of one
thing or half of another since halfs tend to fall apart and get
really messy.
> > It would be cool, too, if we could use namespaces... the
> > UT_ stuff is not very sexy.
Then you would have to do some serious educating to make sure that
all of the developers who might contribute to this project properly
understand the ramifications of the language features that they are
using. This might make the entry level for project participation
prohibitive, or at the least a little more daunting for "beginners".
Just as a case in point look at how many people are having a hard
time keeping the proper use of const in the project.
> Amen! That would be progress in the right direction. It would also draw a
> distinct border between XAP and AP stuff (don't know the difference? Hardly
> do I from time to time :-) ).
This I whole-heartedly agree with. I think that part of
the problem here is that people have begun to forget the
fact that AbiWord is only one component of what was intended
to be a partial suite of applications.
Thomas
-------------------------------------------------------------
Thomas (toe-mah) Fletcher QNX Software Systems
thomasf@qnx.com Neutrino Development Group
(613)-591-0931 http://www.qnx.com/~thomasf
This archive was generated by hypermail 2b25 : Thu Feb 01 2001 - 10:59:44 CST