BASIC anomaly or is it me?

Hi, this is a section of code I’ve been working on and certain bit has me stumped!

Following dropping a file in a window, I get a usermessage received in workspace%
The line at 559 is very intermittent when it chooses to work or give an empty string and yet reading the same block a byte at a time works.
The last two lines just update text on a debug window I’ve tried this on RO4.02 with RPCEmu and with an RPi running RO5.21(15Jan14) and get the same problem.
I have had several copies of the same program running at the same and one might work when the other four don’t.
Am I doing anything wrong?

     559 :   filepath2$=$(workspace%+44)
     560 :   REM byte bang as above doesn't work every time!
     561 :   filepath$=""
     562 :   temp1%=0
     563 :   temp2%=workspace%?44
     564 :   WHILE temp2%<>0 AND temp2%<>13
     565 :     filepath$+=CHR$temp2%
     566 :     temp1%+=1:temp2%=workspace%?(44+temp1%)
     567 :   ENDWHILE
     568 : 
     569 :   PROCseticontext (debug%,11,filepath$)
     570 :   PROCseticontext (debug%,12,filepath2$)

*Edit I’ve Wimp_Initialiased the task with version 310 as those are the PRMs I’m working from.

Probably the terminating character isn’t what the $ operator recognizes? Have a look at this:

http://newsgroups.derkeiler.com/Archive/Comp/comp.sys.acorn.programmer/2010-04/msg00020.html

Ok, thanks for that.

599 :  SYS"XOS_GenerateError",workspace%+44TOfilepath2$

When I was looking deeper into the bytes, the filepath string returned in workspace% was always terminated with a 0. I just couldn’t work out why it would work some of the time.

No problem – glad that fixed it! ;)

In the days of the BBC B, strings were terminated by CR. After the Archimedes got going the importance of conformity with C, where strings are terminated by NUL, became more relevant. I think this is what explains much of the inconsistency about string-terminators in the RISC OS PRMs; though others may have more accurate memories than mine.

It is for this reason that RiscLua offers a variety of ways of extracting strings from memory. Thus:

$[x] – NUL-terminated string at x
$(x) – CTRL-character terminated string at x
$(x,n) – string of length n at x

Gavin, while you’re here and I’m here, I have some questions about RiscLua that I emailed over to you the other day. Did you receive them? If not, pls email me on willblatchley [email symbol] yahoo [full-stop] com. Thanks!

I think this is what explains much of the inconsistency about string-terminators in the RISC OS PRMs

String termination is quite a mess. NULL predominates, but LF is used in the Toolbox and just string length is used by some SWIs. And supplied strings can sometimes be control character terminated, but others must be NULL terminated. We will have to live with it, though.

ways of extracting strings from memory.

Basalt has the RETURN$ keyword:

 =RETURN$(x%)      - CTRL character terminated
 =RETURN$(x%,end%) - address of end of string >&8000
 =RETURN$(x%,len%) - length of string <&8000