Re: [AD] 4.2.0 release failed build on VC 2005

[Evert Glebbeek <eglebbk@xxxxxxxxxx>]

On Saturday 31 December 2005 10:21, aj wrote:
> Win98 is still supported, an OS that is, as of tommorow, 8 years old...  

Also an OS that is still used by people who don't have the money to spend 
to buy a newer version of Windows. Also an OS where it takes almost zero 
effort to maintain support for in addition to other Windows ports.

> Do you expect the same of linux?   Would you limit the advancement of  
> allegro because someone wanted to run a 8 year old linux distro, that 
> had some quirky behaviour ?

Upgrading Linux versions doesn't cost the same as upgrading Windows 
versions. Either way, this wasn't about what versions of Windows to 
support and not to support, it was about compilers.

> This isn't the same argument for dropping DOS support, DOS is a seperate 
> platform, this isn't an argument about platform, its about compiler 
support.

Which is exactly why I didn't bring it up. And by the same token, Windows 
versions have nothing to do with any of this either.

>  >>what does dropping MSVC6 buy us in that case?
> 
> i think it buys us the ability to spend limited developer time on 
> developping forward, instead of backward.

I spend zero time on MSVC support of whichever version.

> i think it buys us the freedom to add features in v7 and v8 that are not 
> available in v6... closer to C99 support.

Neither MSVC 7 nor 8 fully supports C99. Which makes the C99 support fairly 
useless in my opinion. Either we say `Allegro is C89, use a compiler that 
is C89 complient' or we say `Allegro is C99, you need a C99 compiler'. We 
should not say `Allegro is mostly C89, but has some C99-isms in it, so you 
should use a compiler that sortof not quite supports C99'.

> i think it buys us a better allegro, a more stable allegro, a faster, 
> feature richer allegro.

How? How does making Allegro not compile with MSVC 6 help with any of that? 
Remember, using a new tool does not magically improve the product you make 
with that tool.
I'm not saying actively supporting new compilers is a bad idea or that we 
should not have ways to take advantage of them (and I haven't seen anyone 
say that), but you simply shouldn't deliberately break something that you 
don't have to break.

Evert