Re: [AD] MSVC 8 cl flags

[Matthew Leverton <meffer@xxxxxxxxxx>]

> > What I think we should drop for after 4.2.0 is things like defining all
> > variables at the beginning of blocks - I find this is just bad style and
> > leads to programming errors.
> 
> This is a very volatile discussion point, and my point of view follows Owen
>
Regardless of which way to declare variables is better, dropping
support for MSVC 6 or later would be quite foolish. Microsoft is
making a big push with their "Express" version, and it is infinitely
better than any open source or free offering for Windows when it comes
to the IDE and built-in debugger. I suggest you download it and test
it out. I'm very impressed with it, and would even buy it if the final
retail price is under $100.

MSVC's popularity with new programmers will only increase, and
refusing to acknowledge that is a big mistake. On a.cc, MSVC builds of
Allegro are downloaded almost 3 times as much as MinGW builds and
about 1 to 1 with source versions. And so even if you unfairly grant
MinGW all the source downloads, MSVC users still account for around
40% of all downloads.

Also, MSVC 6 is still widely in use. And until the upgrade path to
Express editions is clear, it probably still will be for a while. I
think going forward with 4.3, if MSVC 6 is majorly holding Allegro
back then it would be fine to drop it. In any case, I really don't see
how MSVC 6 holds back Allegro anymore than Watcom, BCC32 or DOS
support does. Switching to a supposedly better coding style or
refusing to add a few hacks is not worth losing a good portion of your
user base. It's not good to get trigger happy just because you don't
happen to use something.

When MSVC stops compiling standard C applications, then it should no
longer be considered as an official target. Until then, it should
remain a top target for Allegro, because--like it or not--it's
probably the most popular IDE for Windows.

--
Matthew Leverton