Waldek Hebisch <hebisch <at> math.uni.wroc.pl>
2016-03-29 02:22:23 GMT
Paul Isaacs wrote:
> As it is I was thinking function-access would be accessing a function,
> not a return value.
> Component-function-access with array and records was providing clues but
> I just couldn't figure out the meaning.
> Am I correct in now thinking that 10206 will not provide any way to get
> the address of a function the way that the <at> extension does in gnu
> pascal without the --extended-pascal restriction?
In ISO Pascal you have variable parameters and procedural parameters.
They allow you to do most things possible with address operator,
but not all and in a different way.
> I am chugging away writing a 10206 compliant compiler using GNU pascal
> and suddenly thought that using extensions to enable the writing of a
> compliant compiler was not a great way to advertise 10206. I recognize
> that the extensions are there because they are useful - but - do they
> also flag 10206 as a poor standard to use?
Well, 10206 was deliberately limited. Practically, decent compiler
should allow writing runtime library. But ISO 10206 misses
extentions needed for writing functions like 'new'. This
does not mean that this is poor standard, just that for
some purposes you may need extentions.
> I am hesitating to use extensions because the target of the compiler
> will be embedded processors.They are problematic to debug so you want to
> have the compiler produce code that is as bug free as possible. I though