A89: Re: Re: Re: nostub programs and shells.
[Prev][Next][Index][Thread]
A89: Re: Re: Re: nostub programs and shells.
Hi!
> Actually, they think that the only nostub programs that can
> (should?) exist are kernel installers. So, they prevent nostub
> to run to avoid reinstalling the kernel by error.
I think that every kernel installer must have a signature in itself,
so the shell should to use this signature to recognize what is the
kernel installer. But, so far, there is no conviences about how the
correct program for TI should to be organized. This is a pity.
> Also, many programmers (not me) hate nostub programs because :
> - the relocation table make them bigger
My experience shows that 40K long program compiled in "nostub" mode
becomes 38-39K long in "DoorsOS" mode, so this is not so drastical
difference.
> they have to embed many routines usually provided by libraries or
> ramcalls, so they won't benefict from improvements in futures
> librairies versions.
This fact is mainly true.
> they are supposed to be less compatibles with futures rom versions.
Absolutely incorrect.
> (In fact, nostub programs are often the first ones to run on new AMS
> versions)
Exactly.
> That might also explain why shells are not going to run nostub programs
> for a long time...
Yes.
Cheers,
Zeljko Juric
Follow-Ups: