Re: philosophical argument / width of int for gcc targets

["Björn Haase" <Bjoern dot M dot Haase at web dot de>]

"Björn Haase" <Bjoern.M.Haase@web.de> schrieb am 02.01.05 19:10:10:
> 
> 
> Robert Dewar <dewar@adacore.com> schrieb am 02.01.05 17:52:33:
> > 
> > Paul Schlie wrote:
> > 
> > > Well considering the fact that most of the worlds processor consumption (by
> > > orders of magnitude), are based, and likely will continue to be based 16-bit
> > > or less processor architectures, which not only care about 16-bit or less
> > > ints and pointers, but rely on them; it should likely be considered
> > > sufficiently significant to warrant concern. (although fortunately, most
> > > optimizations related to overcoming inefficiencies of deeply pipelined
> > > larger machines are largely irrelevant on smaller lightly pipelined
> > > architectures with closely coupled memory sub-systems).
> > 
> > This is a little misleading, yes, most of the worlds processors are 16-bit
> > or less, but if you count programs, you will find that 32-bit processors
> > dominate by a huge margin. That's because the typical pattern is a million
> > watches with the same (rather small and simple) program.
> > > 
> > 
> > 
> 
 In my opinion it's not useful to start philosophical arguments about how
 important which kind of platform is. Anyone knows that both have their
 justification and I understand both sides. In my opinion the key problem is
 
 1.) Most people that are skilled enough to understand the gcc code and 
 develop it further never ever design code for 8 or 16 bit target systems. 
 I can therefore understand that they do not recognize immediately that there
 is a big interest in free software also for smaller targets.
 2.) I understand that some people might think that "carrying with them" the 
 small targets is not useful at all.
 3.) On the other hand, there is another community, mainly people with an 
 engineering background, that is working with 8/16/24 bit systems and that is
 interrested in using free software. Mostly people at education institutions 
 and people working in small companies that whish to escape the
 extreme costs for sw development tools for embedded HW and whish to 
 avoid to deliver themselves to big SW companies that keep their sources
 secret. It is possible that most gcc developers never ever meet one of 
 those people. But in my opinion one should not deny that this 
 community exists also.
 
 In my opinion the key question is whether the possible burden is so 
 large that it does not justify maintaining portability. My personal impression 
 is: no. (Of course keep in mind, that my personal view is biased since I am 
 interested in the avr target and it's not me who will encounter
 the difficulties.) Firstly, I think that clearly maintaining separate target 
 and host representations for data objects might be good for itself. Having 
 targets with different architecture than the host forces one to maintain this. 
 And given that, I have the hope that the remaining amount of work that is
 required for continuing to support small target architectures is not so big 
 that it would justify to deliver people working with small targets to the 
 commercial SW industry. 
 
 Yours,
 
 Björn
 
 P.S.:
 It is true that there is this sentence claiming
 "gcc aims to be portable to any architecture where int is 32 bit" if I recall
 properly. I personally, however, understood it always so that this sentence 
 refers to the host platform gcc is running on and not to the target system.


__________________________________________________________
Mit WEB.DE FreePhone mit hoechster Qualitaet ab 0 Ct./Min.
weltweit telefonieren! http://freephone.web.de/?mc=021201