Auto GEM code generation from RSC in FORTH

[Moulinaie]

It does not seem to be caused by autoexec.for. I get that message regardless whether the file exists or not.

Maybe try to run it in the same scenario in Hatari, and trace gemdos calls will help.

BTW, typing "exit" without anything on the stack gives a bus-error.

exit does not exit.
Well, actually yes, but within a word to come back to the parent word.
If you want to get out to the desktop, use "system".

Guillaume.

[ThorstenOtto]

Yeah, thats clear. But still it should not cause a bus-error

[ThorstenOtto]

The "bad Fopen" from Hatari is caused by an Fopen with an empty filename. Examining the memory from the logs above verifies that:

Code: Select all

> memdump 0x10E52CF
010E52CF: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00   ................
010E52DF: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00   ................

[ThorstenOtto]

The error dialog disappears when i tell gemini *not* to use ARGV for GEM-Programs. ARGV means that the first byte in the commandline is 127. I think FORTH does not handle that, and tries to interpret the remaining commandline (which is normally empty, since the actual parameter are passed in the ARGV enviroment variable) as a filename.

[wongck]

What RSC editor did you use?

My own of course (ORCS)

I was like... isn't it obvious??? His own, of course.

[Moulinaie]

The error dialog disappears when i tell gemini *not* to use ARGV for GEM-Programs. ARGV means that the first byte in the commandline is 127. I think FORTH does not handle that, and tries to interpret the remaining commandline (which is normally empty, since the actual parameter are passed in the ARGV enviroment variable) as a filename.

Thanks a lot!
You're right I don't handle this. I will have to modify that.

Guillaume.

[ThorstenOtto]

If you are looking for assembler code to handle ARGV parsing, you can take a look at https://github.com/th-otto/MagicMac/blo ... /cstartv.s or the pcvstart.s source of Pure-C. Both versions construct an array of pointers to the arguments at the bottom of the heap.

[Moulinaie]

If you are looking for assembler code to handle ARGV parsing, you can take a look at https://github.com/th-otto/MagicMac/blo ... /cstartv.s or the pcvstart.s source of Pure-C. Both versions construct an array of pointers to the arguments at the bottom of the heap.

Thanks!
I have made rapidly a partial support for it. Sometimes in the future I will implement it all.

I have uploaded the new version.
Guillaume

[Moulinaie]

Yeah, thats clear. But still it should not cause a bus-error

This is fixed too!

Guillaume;

[ThorstenOtto]

Confirmed. Passing parameters works now, and the error dialog is gone. The bus-error is also gone.Thank you!

Just another thing i noticed: i think you should locate the forth.inf file with shel_find(). Otherwise, if you assign FORTH.PRG to eg. *.for, then start it by double-licking a file that is not located in its directory, and have set the desktop settings such that the default directory is "Window" (and not application), it won't be found. This happens both with Gemini and the standard desktop.

[Moulinaie]

Confirmed. Passing parameters works now, and the error dialog is gone. The bus-error is also gone.Thank you!

Just another thing i noticed: i think you should locate the forth.inf file with shel_find(). Otherwise, if you assign FORTH.PRG to eg. *.for, then start it by double-licking a file that is not located in its directory, and have set the desktop settings such that the default directory is "Window" (and not application), it won't be found. This happens both with Gemini and the standard desktop.

That's clever!
I'll add it.

Have you noticed that with rsc2for, I have added a summary of the RSC file at the end of the source? You get all defined names with their type and, for some objects, with their flags. That's a reminder you can use to know who's who when modifying the text.

Guillaume.

[Moulinaie]

Confirmed. Passing parameters works now, and the error dialog is gone. The bus-error is also gone.Thank you!

Just another thing i noticed: i think you should locate the forth.inf file with shel_find(). Otherwise, if you assign FORTH.PRG to eg. *.for, then start it by double-licking a file that is not located in its directory, and have set the desktop settings such that the default directory is "Window" (and not application), it won't be found. This happens both with Gemini and the standard desktop.

That's clever!
I'll add it.

Have you noticed that with rsc2for, I have added a summary of the RSC file at the end of the source? You get all defined names with their type and, for some objects, with their flags. That's a reminder you can use to know who's who when modifying the text.

Guillaume.

EDIT: shel_find done !

[ThorstenOtto]

That was fast Yes, well done, seems to work.

[Moulinaie]

That was fast Yes, well done, seems to work.

I did that for M_Player but never had the idea to transfer it to Forth...

Guillaume.