From: Betov on
Frank Kotler <fbkotler(a)verizon.net> �crivait
news:XLpDi.4437$Ov2.916(a)trndny06:

> I'm not sure what the "problem" is.

In your code, there is, for example:

jmp 'libgtk-win32-2.0-0.gtk_init'

RosAsm is unable to re-Assemble this, after a Disassembly,
because it cannot guess which '.' is the separator in
between the ModuleName and the FunctionName. In this case,
it supposes that ".0-0" is the Module Extension. :)))))))

I will have to implement another mechanism. Anyway, for
GTK programming, we are not going to like typing all that
awfull stuff in...

I suppose that the simpliest solution will be to create
a special Declaration saying:

[IMPORT: 'libgtk-win32-2.0-0', 0
'gtk_init', 0
... ]

.... or something like this...


> I would have expected that a bigger "problem" to you would have been
> using gcc to link 'em.

No. If i would have to do that, i wouldn't do anything.
How do i do it under Win32?


> I have *no* idea what that "'pkg-config...'"
> bit on the command line does. Calling just "printf" (libc) or Xlib, I
> can get an executable out of ld, without going through gcc and linking
> in the "startup code". I don't know how "robust" that is, but it
> "seems to work". There may be a way to do it for gtk, too, but I
> haven't found it.

Yes, it was quite a pain for me to rebuild the FASM/GCC/GTK
examples. Several days at searching for what is what in the
Linux+C hell. 100,000% unusable.


> http://www.mpeforth.com/arena/cwgtLinux.txt
>
> Don't be put off by the fact that it's "about" writing a Forth
> compiler. It shows how to get dynamically linked code out of Nasm
> without a linker. You may need to get "grounded" in the ELF format to
> understand it... as I said, I haven't tried it. It suggests that code
> that calls shared libraries *can* be produced by RosAsm/Wine...
> maybe...

Yes, This is what i will begin to try soon.


> I hope you stick with Linulong enough to do something with it, but
> if it continues to be painful (I don't know if "three weeks" is too
> long or too short), don't keep beating your head against it. Life's
> too short!

Well, nothing new. How do you think i analysed the PE Format
ten years ago, whereas the Docs were not even complete? Yes,
by analysing a PE Byte after Byte, until i could rebuild one.


Betov.

< http://rosasm.org >




From: Frank Kotler on
Betov wrote:
> Frank Kotler <fbkotler(a)verizon.net> �crivait
> news:XLpDi.4437$Ov2.916(a)trndny06:
>
>
>>I'm not sure what the "problem" is.
>
>
> In your code,

Not my code, Numit_or's code.

> there is, for example:
>
> jmp 'libgtk-win32-2.0-0.gtk_init'

I don't see that construct in Numit_or's code.

> RosAsm is unable to re-Assemble this, after a Disassembly,
> because it cannot guess which '.' is the separator in
> between the ModuleName and the FunctionName. In this case,
> it supposes that ".0-0" is the Module Extension. :)))))))

Mmmm... where the actual extension is ".dll"? (since you seem to be
looking at the Windows code...) What Linux sometimes does is to put a
"symlink" file in /lib libfoo-0.0.1.so.2 -> libfoo.so or whatever. That
might help you, but would mean that your users would have to create such
a symlink, too.

> I will have to implement another mechanism. Anyway, for
> GTK programming, we are not going to like typing all that
> awfull stuff in...

"int 80h" is definitely easier to type. :)

> I suppose that the simpliest solution will be to create
> a special Declaration saying:
>
> [IMPORT: 'libgtk-win32-2.0-0', 0
> 'gtk_init', 0
> ... ]
>
> ... or something like this...

I think there's a way in the Nasm "import" section (only in "-f obj"
output format!) where you can do:

import my_beautiful_name foo.dll TheirUglyName

Something like that?

>>I would have expected that a bigger "problem" to you would have been
>>using gcc to link 'em.
>
>
> No. If i would have to do that, i wouldn't do anything.
> How do i do it under Win32?

Hmmm, I guess the Windows versions use just ld, not gcc. I wonder if I
can learn from that.... Lessee... wouldn't have to tell Linux ld about
subsystems - there's only one. How an you program in assembly when
there's no one Windows, but all these different subsystems? :) "$LPATH"
and "$LIBS"... looks the same as what we'd tell Linux ld...

>>I have *no* idea what that "'pkg-config...'"

No, no hint there...

....
> Linux+C hell. 100,000% unusable.

Yet people use it every day...

Best,
Frank
From: //o//annabee on
P� Wed, 05 Sep 2007 11:41:47 +0100, skrev Frank Kotler
<fbkotler(a)verizon.net>:

>> RosAsm is unable to re-Assemble this, after a Disassembly,
>> because it cannot guess which '.' is the separator in
>> between the ModuleName and the FunctionName. In this case,
>> it supposes that ".0-0" is the Module Extension. :)))))))
>
> Mmmm... where the actual extension is ".dll"? (since you seem to be
> looking at the Windows code...) What Linux sometimes does is to put a
> "symlink" file in /lib libfoo-0.0.1.so.2 -> libfoo.so or whatever. That
> might help you, but would mean that your users would have to create such
> a symlink, too.

Could not the tool or installer do it for them?

> Best,
> Frank

From: CodeMonk on
Frank Kotler wrote:
>
> Absolutely!!! Do you realize that doctors peddle drugs they haven't even
> *tried*?
>

Good, I wouldn't want anyone to confuse misinformation with plausible
deniability in reverse. Do they have a program of misconduct medals
for those doctors? Surely they wouldn't peddle drugs that haven't
been through clinical trials.

One would never want the patients to be misinformed. Unless of
course, if one is to believe the goals of such meds is to silence the
rational and amplify the delusional. I'm not especially good with
medical euphemisms, but I understand the bandpass filtering concept.

You said:

"Hutch left one out:"
"I can heal the sick, I can raise the dead"

Hutch actually said:

"Its too late to heal the sick and
there is no chance of raising the dead"

So other than being an apparent contextomy, your point is?

Then there's also this ...

You said (many moons ago):

"I can make the little women
talk out of their heads."

Which you rephrased for me as:

"I can make your little woman
talk all out of her head"

"But, no... that's not me."

And yet, so it was, and therefore, so it is.

- Scott

PSS: What, no Beverly Hillbillies Episodes today?
They're almost heavenly.
From: Wolfgang Kern on

"Betov" replied:

>> Now all what I found were Libraries and tool offers,
>> the page named "API-reference" seems to be a very blank one.

> ? Did you click on [Api References], in the Documentation
> menu ? ---> < http://www.gtk.org/api/ >. Then you click
> upon, say, "online", for "Glib"... and so on. Tons of stuff
> there. I did not really digged into, but i found out several
> details, i am concerned with. Nothing much different from
> the Win32 Api we are used to, in my opinion (to be verified).

Yes all what I found were tons of LIBs. And I dont want to scan
them all just to get a list of linux funtions.

> To me, the problem is not a this level: It is *below*. The
> problem is *Linux*. This system is utterly unusable. Scaring.

Looks like.
I got an Xubuntu CD, it needs half an hour to boot and then asks:
"install ?", and waste you another 20 minutes if you say yes.
And finally this OS is a DO-goaNIX (xlat: do nothing at all).

>> Where the hell are the description of this damned linux functions
>> from ADD_WCH until ZCAT to be found ???

> Don't ask me... :)))))

> But if they are *Linux* Functions, i suppose they are not
> in there, unless the GTK stuff would cover them (I saw GTK
> funtions for Files, for example. So...)

Regardless if they are functions/structs/or whatsoever
they aren't of any use without information.

>> Use they all just C-libraries and nobody cares how and why they work ?
> Does something "works" under Linux ?> :))

Not yet if you ask me.

>> Someone must have written this libs, Mr. Linus himself ?
> :]]]]]]
> Have a phone ?
> :]]]]]]
> Courage. Betov.

Thanks, I better give this idea a break and wait for someone can
at least point to the proper source of required informations.

__
wolfgang