I'm finding that C:rx doesn't seem to be working quite right in update 4.

eg

rx rexx:dme.rexx filename

used to pass the arg filename to script rexx:dme.rexx in update 3 and before but
now the arg is not passed.

but

rx 'ADDRESS REXX "rexx:dme.rexx filename"'

does work.

Last edited by broadblues on 09-Feb-2006 at 07:45 PM.

_________________
BroadBlues On Blues BroadBlues On Amiga Walker Broad

@broadblues

I also have many pb with upd4 and ARexx. Most of my ARexx
scripts doesn't work anymore (no more autobackup, yups :(.
There is a big pb with arg passing, as you noted.

Arg passing in AWeb & ClassAction is not working, also.
Thinking of downgrading to some upd3 files.

Bye,
TMTisFree

_________________
The engineering approach to our non-problems: "build a better washer".
The scientific approach to our non-problems: "find a new energy source".
The environmentalist approach to our non-problems: "stop washing your shirts".

@broadblues

Quote:

broadblues wrote: I'm finding that C:rx doesn't seem to be working quite right in update 4. eg rx rexx:dme.rexx filename used to pass the arg filename to script rexx:dme.rexx in update 3 and before but now the arg is not passed. but rx 'ADDRESS REXX "rexx:dme.rexx filename"' does work.

Ouch. I rewrote RX from scratch for update #4. Exactly which version of the RX command are you using?

@olsen

same problem here... i just "version rx"

8/0.RAM Disk:> version Rx full
rx 51.9 (01/23/2006)


but writing that 'address rexx' worked for me....

Last edited by Haymigga on 09-Feb-2006 at 08:57 PM.

_________________
0 % Amiga.
www.amigabounty.net

@olsen

Quote:

Ouch. I rewrote RX from scratch for update #4. Exactly which version of the RX command are you using

rx 51.9 (23/01/2006)

_________________
BroadBlues On Blues BroadBlues On Amiga Walker Broad

@TMTisFree

Quote:

Arg passing in AWeb & ClassAction is not working, also. Thinking of downgrading to some upd3 files.

I haven't notice any Arexx related problems in AWeb yet, what script did you try to run and how?

_________________
BroadBlues On Blues BroadBlues On Amiga Walker Broad

@broadblues

Quote:

broadblues wrote: @olsen Quote: Ouch. I rewrote RX from scratch for update #4. Exactly which version of the RX command are you using rx 51.9 (23/01/2006)

Oops. That's the one with the bug. I broke the parameter processing somewhere between 51.7 and 51.8, and the bug was eventually fixed in 51.10.

Looks like it didn't make it onto the Update #4 CD. But this should be easy to fix

In the mean time, please try to use the older 68k RX command.

@olsen

Quote:

In the mean time, please try to use the older 68k RX command.

does that go for RXC RXLIB & RXSET as well?

_________________
BroadBlues On Blues BroadBlues On Amiga Walker Broad

@broadblues

Quote:

broadblues wrote: @olsen Quote: In the mean time, please try to use the older 68k RX command. does that go for RXC RXLIB & RXSET as well?

I am not aware of any problems concerning these commands. While I also rewrote these from scratch, they are a lot less complex than the RX command. Well, RXLIB and RXSET are more complex than, for example, RXC or HI, but they all are very simple in design and implementation, compared to what RX has to do.

So, in a nutshell, the remainder of these commands (RXC, RXLIB, RXSET, HI, TCC, TCO, WaitForPort) ought to work correctly.

@broadblues

Not related with ARexx but with arg passing. For example
in AWeb, when cliking on a jpg link, AWeb complains not finding
Multiview: but Multiview is present and args for managing
jpg images did not change in the meantime.
I have the very same pb in ClassAction: no one of the variables
pass their content.

@olsen
Is it possible to get the right version (I mean bug-free :)
of RX for testing (and more if affinity :)

Bye,
TMTisFree

_________________
The engineering approach to our non-problems: "build a better washer".
The scientific approach to our non-problems: "find a new energy source".
The environmentalist approach to our non-problems: "stop washing your shirts".

@olsen

All pb solved with RX 68k, thanks. So it appears that args
passing in ClassAction and AWeb requires RX to work, good to
know.

Bye,
TMTisFree

_________________
The engineering approach to our non-problems: "build a better washer".
The scientific approach to our non-problems: "find a new energy source".
The environmentalist approach to our non-problems: "stop washing your shirts".

@olsen

RXSET also does not pass parameters correctly.

Example:

9335 *-* If Word(ShellResult('`version workbench.library`'),2) ...
>>> "`version workbench.library`"
12196 *-* ShellResult:
12196 *-* ;
12197 *-* Parse ARG ShellCmd;
>>> "`version workbench.library`"
12198 *-* cmdid = 'req'Pragma('ID');
>>> "ID"
>>> "req01A8A160"
12199 *-* Address COMMAND 'rxset' cmdid ShellCmd;
>>> "rxset req01A8A160 `version workbench.library`"
12200 *-* from_shell = GetClip(cmdid);
>>> "req01A8A160"
>>> "workbench.library"
12201 *-* Call SetClip(cmdid,'');
>>> "req01A8A160"
>>> "" -------------------------------- returns nothing (has to be the version of the workbench.library)
>>> "1"
12202 *-* Return(from_shell);
>>> "workbench.library"
>>> "workbench.library"
>>> "2"
>>> "1"
9335 *-* Call _ReqErrorInit("CSVControl required AmigaOS v3.5+...
>>> "CSVControl required AmigaOS v3.5+."
>>> "Exit"

Have to switch to the good'old 68k, it seems.

Bye,
TMTisFree

_________________
The engineering approach to our non-problems: "build a better washer".
The scientific approach to our non-problems: "find a new energy source".
The environmentalist approach to our non-problems: "stop washing your shirts".

@TMTisFree

Could you write up a complete command line that doesn't work for me? I'm not sure if what I think might be the right line actually produces the problem. So rather than chasing a bug that came together because I couldn't read correctly from the screen, I'd rather see what you found

@olsen

Yes sorry this debug output also gives me headheaches
sometimes...

Save below in a file and RX it:


/*the line requesting the version of the workbench.library*/

Say Word(ShellResult('`version workbench.library`'), 2)
Exit

/*the function returning the result of the shell command*/

ShellResult:
Parse ARG ShellCmd
cmdid = 'req'Pragma('ID')
Address COMMAND 'rxset' cmdid ShellCmd
from_shell = GetClip(cmdid)
Call SetClip(cmdid, '')
Return(from_shell)





My previous post shows that rxset fails to returns the version of worbench.library.
The RXset 68k returns correctly the version.

Hopes this helps

Bye,
TMTisFree

Last edited by TMTisFree on 11-Feb-2006 at 10:27 AM.

_________________
The engineering approach to our non-problems: "build a better washer".
The scientific approach to our non-problems: "find a new energy source".
The environmentalist approach to our non-problems: "stop washing your shirts".

@TMTisFree

Quote:

TMTisFree wrote: @olsen Yes sorry this debug output also gives me headheaches sometimes... Save below in a file and RX it: [..] My previous post shows that rxset fails to returns the version of worbench.library. The RXset 68k returns correctly the version. Hopes this helps Bye, TMTisFree

Thank you, I can reproduce the problem. The problem is in the parameter that is passed as the "value" of the clip list entry. That parameter is "workbench.library 51.xxx", but the parser in RXSET only uses the "workbench.library" part and drops the rest. This is because the command line processing works differently in the rewritten RXSET command.

I don't want to bore anybody with details, but a general word on how the original ARexx shell commands worked might be in order These commands were very small and very compact. They did what they had to do with a minimum of code and delegated all the complex work to rexxsyslib.library functions. A common trait of these shell commands is in that the command line processing is somewhat simple. The command line contents are offered as one single line of text to a shell command, and the ARexx shell commands had to break this down into individual parameters. There are more than just a few side-effects hidden in there. One of them is, for example, with the RXSET command, which accepts the "=" character at any place in the command line string, e.g. "name=value" is equivalent to "name= value", "name =value" and "name = value". This does not work with the default AmigaDOS shell command parameter processing, which can choke on the "=" if put in the wrong place. This is why the RXSET rewrite tries to be clever, but is not clever enough. I'll try to do something about it.

@olsen

Technical explanations are always welcome, thanks for that.

About the parsing pb, keeping the 2 parsing methods might
keep compatibility without adding much lagging or weight to
the ARexx commands.

I see rexxsyslib.library is now native. Do you also plan
to convert the small rexxsupport.library?

Edit:typos

Bye,
TMTisFree

Last edited by TMTisFree on 12-Feb-2006 at 11:15 AM.

_________________
The engineering approach to our non-problems: "build a better washer".
The scientific approach to our non-problems: "find a new energy source".
The environmentalist approach to our non-problems: "stop washing your shirts".

@TMTisFree

Huh? Rexxsyslib is not native. A glue code file (rexxsyslib.l.main) goes with it.

_________________
Philippe 'Elwood' Ferrucci
Sam460 1.10 Ghz
AmigaOS 4 betatester
Amiga Translator Organisation

@TMTisFree

Quote:

TMTisFree wrote: @olsen Technical explanations are always welcome, thanks for that. About the parsing pb, keeping the 2 parsing methods might keep compatibility without adding much lagging or weight to the ARexx commands.

The parsing is a nightmare. On the one hand you want to keep compatibility high, on the other hand you want to unify the processing of the command line parameters throughout all the shell commands. Turns out that there are many side-effects hidden in the way the original ARexx shell commands allowed you to use them that you are bound to miss something in a reimplementation. You can't catch all the weird cases.

RX was such a nightmare because you could get creative with the use of a double quote character following the command name itself. That double quote could either mean that the name of the rexx script would contain blank spaces, or it could mean that the remainder of the line would contain a single compact rexx program.

Quote:

I see rexxsyslib.library is now native. Do you also plan to convert the small rexxsupport.library? Edit:typos Bye, TMTisFree

Sorry, rexxsyslib.library is still a single lump of 68k code. It has to be ported, eventually, but we're not quite there yet.

@elwood

Ah so my mistake. I thought that rexxsyslib.l.main
was the native counterpart of rexxsyslib.library.

@olsen

Unification is probably a good thing. And it is not that
difficult to maintain/modify ARexx script, even old one
if needed. Just don't forget to add some documentation on
how to respect this new parsing.

Bye,
TMTisFree

_________________
The engineering approach to our non-problems: "build a better washer".
The scientific approach to our non-problems: "find a new energy source".
The environmentalist approach to our non-problems: "stop washing your shirts".

@olsen

Has the RANDOM() function been improved ?

Actually (at least on OS3.5), this function is not very reliable as the range of random numbers returned is not well distributed. E.g.: with RANDOM(1,50) the "1" and "50" are never picked ("48", "49" are also rarely picked).

I know RANDU() gives better results, but RANDOM() - if fixed - can save some calculations in simple scripts.