Page 1 of 1 1
Topic Options
#78828 - 2002-05-15 02:40 AM 4.10 - critical script filespec
Anonymous
Unregistered


I've alreadyt alerted Ruud to this problem.

Just in case my testing is flawed, I'd like to solicit some feedback from others.

Try the following:

place a copy of wkix32.exe in one folder.
place your script in a different folder.

attempt to launch as follows:

c:\folder1\wkix32.exe c:\folder2\myscript.kix

In my testing, kix hangs and pegs the CPU at 100%.

I believe it has something to do with the script filespec containing a drive letter, because if I do this:

c:\folder1\wkix32.exe \folder2\myscript.kix

it works. Can anyone else confirm this or are all the test computers on my network acting exceptionally wierd?

Top
#78829 - 2002-05-15 02:58 AM Re: 4.10 - critical script filespec
Shawn Administrator Offline
Administrator
*****

Registered: 1999-08-13
Posts: 8611
Brian,

hmmm, I can't seem to repro that under Windows XP, i copied wkix32 4.10rc1 to f:\folder1 and test.kix to f:\folder2 and ran:

f:\folder1\wkix32.exe f:\folder2\test.kix

i ran a blank script (break on/exit) and a script that just wrote a line to the console - no problems ... what kinda script you running / you using any of those new 4.10 macros ?

-Shawn

Top
#78830 - 2002-05-15 03:11 AM Re: 4.10 - critical script filespec
Les Offline
KiX Master
*****

Registered: 2001-06-11
Posts: 12734
Loc: fortfrances.on.ca
Brian,
I tested as well calling WKIX32.EXE from C:\

C:\>C:\KIX4.1RC1\WKIX32.EXE C:\KiX4.02\test3.kix

Everything ran well. Could not reprodudce your issue.

Platform: Win2k Pro, SP2, SRP1
_________________________
Give a man a fish and he will be back for more. Slap him with a fish and he will go away forever.

Top
#78831 - 2002-05-15 07:19 AM Re: 4.10 - critical script filespec
Jochen Administrator Offline
KiX Supporter
*****

Registered: 2000-03-17
Posts: 6380
Loc: Stuttgart, Germany
and on NT SP6a here also failed to reproduce ...

Jochen
_________________________



Top
#78832 - 2002-05-15 05:40 PM Re: 4.10 - critical script filespec
Anonymous
Unregistered


THanks for the prompt replies. Ruud too can't reproduce the problem, but I now have a debug verion of 4.10 RC1, so I hope to figure this one out. Talk about weird...
Top
#78833 - 2002-05-15 05:45 PM Re: 4.10 - critical script filespec
Shawn Administrator Offline
Administrator
*****

Registered: 1999-08-13
Posts: 8611
Brian

Can you post your test script ? Is it just a simple break on exit 1 script ?

-Shawn

Top
#78834 - 2002-05-15 09:24 PM Re: 4.10 - critical script filespec
Anonymous
Unregistered


Actually, the script I am testing with is over 6,000 lines of code.
I'm now going to try a simpler, smaller, test script to see if I get the same results.

Top
#78835 - 2002-05-16 01:20 AM Re: 4.10 - critical script filespec
Anonymous
Unregistered


Super-simple script works as intended.

Ruud is thinking it may be related to pre-processing UDFs. So I cut and pasted about 70 UDFs into the simple script, and still was unable to reporduce the problem.

I'm leaning toward script size at this point? Anyone tested with a script > 6,000 lines?

Testing continues... will post progress when I have some more results...

-Brian

Top
#78836 - 2002-05-16 05:49 AM Re: 4.10 - critical script filespec
Les Offline
KiX Master
*****

Registered: 2001-06-11
Posts: 12734
Loc: fortfrances.on.ca
Just curious... are your UDFs embedded in the >6000 line script?
If not, how do you call them (what path) and do you use a macro?
_________________________
Give a man a fish and he will be back for more. Slap him with a fish and he will go away forever.

Top
#78837 - 2002-05-16 08:53 AM Re: 4.10 - critical script filespec
Anonymous
Unregistered


Yes, the UDFs are embedded into the main code.
i've just received another debug version of KiX, so hope to have some more results to Ruud shortly and more results posted after that.
-Brian

Top
#78838 - 2002-05-16 04:03 PM Re: 4.10 - critical script filespec
Sealeopard Offline
KiX Master
*****

Registered: 2001-04-25
Posts: 11165
Loc: Boston, MA, USA
Might be a long shot, but with 6000 lines of code, how big is the script in KB? Maybe you are reaching some kind of memory limit in KiXtart with regards to reading the file into memory? I have scripts + UDF libraries up to 170KB which haven't yet caused any problems.

Maybe you could try to create a 6000+ line file filled with just comments and see whether this will cause a problem, too?

Alternatively, if you put the UDFs into a separate UDf file and call it in the beginning of the script, does the same CPU utilization occur?

Just trying to list the things I'd to to troubleshoot that particular problem.
_________________________
There are two types of vessels, submarines and targets.

Top
#78839 - 2002-05-17 07:05 PM Re: 4.10 - critical script filespec
Anonymous
Unregistered


Ok, well after a week of brain-racking and a few megs worth of KiXtart trace files send back and forth, we have some positive results:

No, it had nothing to do with funky quotes, scripts size, quantity of UDFs, etc.

The problem, as weird as it still remains, can be attributed to one line of code -- a line which has been in my script since 1998.

Try this for yourself (from a machine you don't mind hanging):

code:
$rc=box(0,0,10,10,GRID)
$rc=box(6,9,20,71,Å) ; the special char is produced by ALT+143
$rc=messagebox('Hello World','Testing',4112,0)
exit 0

Mind you: This works without problem with KiX 4.02, and will work with 4.10 if you create a shortcut such as "e:\test\wkix32.exe" "\temp\testscript.kix" and then use it to launch the 4 lines of code above.

What Ruud and I have been working on, is that the failure is caused by changing the script argument to include a drive letter such as "e:\test\wkix32.exe" "e:\temp\testscript.kix"

I don't know exactly why, and Ruud is leaving town for a few days, so we'll have to wait for his explanation upon his return...

But the interim solution is to wrap the special character used to generate the box in quotes. This, (according to the manual) is the way it should have been done, but since I never ran into a problem before, it was never looked at...

$rc=box(6,9,20,71,"Å")
; Wrapped in quotes, the script filespec with a drive letter now works.

So, while there is still a mystery as to why a change in the command line argument triggers the failure, at least I have a solution.

Of course, this may be the tip of the iceberg and other things may be effected, so hopefully finding this one problem will prevent others [Smile]

-Brian

Top
#78840 - 2002-05-17 07:42 PM Re: 4.10 - critical script filespec
Shawn Administrator Offline
Administrator
*****

Registered: 1999-08-13
Posts: 8611
Nice work Brian!

Guess my little script didn't hang you system then ? [Frown]

-Shawn

Top
#78841 - 2002-05-22 04:37 PM Re: 4.10 - critical script filespec
Anonymous
Unregistered


Ruud is back in town. Here's a reply I just received:

quote:
Well, the brilliant news (this one was seriously getting on my nerves) is that I've repro'd, found and worked around the issue.

The weird thing is that it turns out to be a really obscure (timing-related) issue in one of the C-runtime functions. Never hit this before, and it's really ugly: doesn't occur when run in the debugger, only when run 'for real' (this is part of why I couldn't repro anything earlier....).

Wow, glad to get rid of this one: thanks for your excellent support!

-Ruud


Top
Page 1 of 1 1


Moderator:  ShaneEP, Arend_, Jochen, Radimus, Glenn Barnas, Allen, Ruud van Velsen, Mart 
Hop to:
Shout Box

Who's Online
0 registered and 1046 anonymous users online.
Newest Members
fgarcia, Viginette, ManuvdWielNL, Sir_Barrington, batdk82
17889 Registered Users

Generated in 0.058 seconds in which 0.023 seconds were spent on a total of 12 queries. Zlib compression enabled.