[A86] Re: ROM / RAM equates database

[Prev][Next][Index][Thread]

[A86] Re: ROM / RAM equates database


> Ah well, your e-mail looks horrible now. With all these '=3Dnn' things
> which should stand for <LF> and the rest. I know that is because of MY
> mailprogram being old...

I noticed that.  It only happens on this list.  The default charset was=20
us-ascii, so I changed it to iso-8859-1.  Hopefully it's fixed now.  If n=
ot=20
I might try Evolution.

> If someone knows a good e-mailer for Windows and not being Outlook
> (Virusses) or Eudora (just to much options), I would be happy to take a
> look at it...

I use Outlook Express, and have never gotten a virus, though YMMV.  Just=20
don't run attachments from people you don't trust, or perhaps never run=20
attachments, and you'll be fine.

If you like console mail readers, you might see if there is a Cygwin=20
version of mutt.  I know a lot of people who really like that.

> Don't say that too loud! That would mean that I would get a LOT of
> troubles in sustaining Venus support... Ever tried a to convince a link=
er
> to add a certain lib at a certain place? (most of them do have an optio=
n
> to take the startupfile first, thank god!)
> It might be that Z80ASM is capable of that though...

Maybe you misunderstood me?  You said that Venus places external variable=
s=20
at the end of the program while it's executing.  Shouldn't you then know=20
exactly where they are?

> Oh, the externals do get a JP/pointer setup at a fixed adress, I could
> use that one (but the lib wouldn't know it)... Basicaly these are
> starting-offsets...

Right, that's what I'm talking about.  The program knows, not the lib.  I=
n=20
many cases that's sufficient.

> See for yourself... http://sdcc.sourceforge.net/ , they also have a
> CVS-tree (for the most current source) and Win/Linux snapshots.

I'll check it out.

> See for yourself, it has Z80, x86, and some other processors. Modula 2,
> C, Pascal, etc. frontends are also present. The pricetag is only a *bit=
*
> high...

In that case, it's worthless.  Is there a demo version available?  If it'=
s=20
decent, I'd like to see how decent.

> "That's not my problem"... :-P

Heh.

> I'd say "Seems non-ANSI to me"...

Sure.  But because you are using inline assembly, it's not portable anywa=
y!=20
I believe a more correct way would be #pragma, but it's much more messy.

> If you know any such language... Your "language X" off coarse, but from
> what I've seen it would be horrible to build a compiler for that (when
> you wnat to implement all the things you stated)

We are still working on the perfect language for the calc :)  Or are we? =
=20
What happened to David West?

Honestly, I'd rather do a language like we are talking about than a C=20
compiler.  It would take some work to get a C compiler just right. =20
Although there is the benefit of having plenty of sources to look at, and=
=20
you don't have to make the grammer yourself.  It takes some time to get a=
=20
grammer correct for a non trivial language.

> You really need to check out one of the Small-C compiler found at
> Ticalc.org. Then you would know that linking is much better than just
> needing included everything. You then most of the time include too much
> code, or you could put everything in dozens of smaller include-files,
> which the programmer should get to know what to use. See Ti8xcc for a
> good example. I have worked with that one for about a year.

No, I still believe I'm right here :)

When it's all included, the compiler can make the best decisions about wh=
at=20
to do with the code.  The optimizer generates flow graphs, so it knows wh=
at=20
is used and what is not used.  Thus it eliminates any dead code.  It gets=
=20
tricky when it comes to function pointers, but there should be a way to=20
handle that as well.

> $nnnn is an adress where "JP vnROMcall" is stored that has been setup b=
y
> Venus... I thought you wanted a solution for the ROM_CALLing on the Ti8=
2
> and Ti85?

Ahh.  Really, I don't care.  I have only seriously coded for the 86.

> Google =3D> "Superoptimizer" (also try changing the 'z' to an 's')
>
> There's some site about some students projects. Another site of the GNU
> superoptimizer (optimizes for the SPARC, MC68000, MC68020, M88000, POWE=
R,
> POWERPC,AM29K, I386, I960 (for i960 1.0), I960B (for I960B 1.1), PYR,
> ALPHA, HPPA and SH), can send you sourcecode of this one. And some site
> about a PIC superoptimizer, have this sourcecode too... Pretty much
> evidence of it's existence, I think...

Interesting.  I will have to check this out.

> The chess game has been 'calculated' before (I've read in "Pythagoras",=
 a
> Dutch maths-magazine)... It is just that cashing (what you litterly do
> then) is tremendous slow when there is that gigantic amount of data to
> scan. I guess the chess game outcomes haven't been analysed completely,
> since THAT would take a long time. I guess that also what you wanted to
> say, analyzing is the main problem.

Btw, that is "caching".  For all practical purposes, there are an infinit=
e=20
number of games.  Although I do believe that due to the rules of chess (n=
o=20
captures in 50 moves, 3 move repetition), there is actually not an infini=
te=20
number of games.

Right.  There is no way to actually compute that many games and store all=
=20
the data for them, at least not presently.

> Where to get that book? "Compilers" is a very general title, I guess
> there are about a dozen of these around.

Sorry, I mentioned the full title above.  "Compilers: Principles,=20
Techniques, and Tools".  Published by Addison-Wesley, ISBN 0-201-10088-6.=
 =20
It's written like a textbook, so it's terse, and you'll likely have to re=
ad=20
it several times to get anything out of it, but it's very good.  I don't=20
know of a better reference on the subject.

> Nice to see that I'm not the only geek who reads that 'much'. :-)
> People at my school bashed [right word?] me because of reading 5+2 book=
s
> of let's say 500 pages. And then I've also read H2G2 (relatively small,
> but dozens of things to 'see'), about all books of David and Leigh
> Eddings, The Hobbit and off coarse, Lord of the Rings... This seems to =
be
> fairly abnormal to most people... :-(

It's very good.  I sure don't read that much, besides technical stuff.  T=
he=20
Hitchhikers Guide to the Galaxy "trilogy" is definitely one of my favorit=
e=20
series, and it gets better everytime you read it.  I also enjoyed the=20
Ender's Game series.  Though, I need to get them all and read them again,=
=20
because when I read them last (7th grade), there were only three books, a=
nd=20
now there are five or six.
References: