LB Booster: Internal Error 15

LB Booster
General >> General Board >> Internal Error 15
http://lbb.conforums.com/index.cgi?board=general&action=display&num=1410279497

Internal Error 15
Post by JLThompson on Sep 9^th, 2014, 4:18pm

My first try at using this resulted in an "Internal Error 15." No explanation anywhere I can find. What is this?

Re: Internal Error 15
Post by Richard Russell on Sep 9^th, 2014, 5:40pm

on Sep 9^th, 2014, 4:18pm, JLThompson wrote:

My first try at using this resulted in an "Internal Error 15." No explanation anywhere I can find. What is this?

That is an internal 'Bad subscript' error. The most likely cause would be that you have exceeded the maximum allowed number of objects of a given type. The limits in LBB are currently set as follows:

Number of program lines: 65535
Number of labels: 10000
Number of FUNCTIONs: 1000
Number of SUBs: 1000
Number of GLOBALs: 1000
Number of STRUCTs: 1000
Number of arrays: 1000
Number of handles: 1000
Number of string variables: 1000
Number of FUNCTION or SUB parameters: 100

Does your program exceed any of these limits? If you don't think so, send me a copy (or a link) so I can try it for myself.

Richard.

Re: Internal Error 15
Post by JLThompson on Sep 9^th, 2014, 5:44pm

I am sure that it exceeds several of these limits. I had hoped that this would save me from having to build a massive network to get the speeds I need. No way I can make my program so small, though, as to let it use this.

Re: Internal Error 15
Post by Richard Russell on Sep 9^th, 2014, 6:42pm

on Sep 9^th, 2014, 5:44pm, JLThompson wrote:

No way I can make my program so small, though, as to let it use this.

Those limits aren't "small"! LBB is used to compile what, as far as I know, is the largest program ever written in Liberty BASIC, Cabinet Planner, with about 100,000 lines in three modules:

http://www.cabinetplanner.com/

The current limits, whilst generous, are in most cases arbitrary so if you need any of them increased just let me know and I will release an updated version.

Richard.

Re: Internal Error 15
Post by JLThompson on Sep 9^th, 2014, 7:05pm

Well, this is an AI experiment 10 years in the writing. It has to be large. I have well over 1,000 string variables, well over 1,000 arrays, could be over 1,000 handles, the lines might be over 65,535 too, I would have to check. Could all of these limits be increased?

Re: Internal Error 15
Post by Richard Russell on Sep 9^th, 2014, 8:02pm

on Sep 9^th, 2014, 7:05pm, JLThompson wrote:

Could all of these limits be increased?

They can all be increased, but there's an issue as far as the total line count is concerned. If an error occurs at run-time the line number is reported, but that can only be a 16-bit number so in the event of there being more than 65535 lines in total the reported line number would be evaluated 'MOD 65536'.

A program of that size is always going to be difficult to maintain. Have you considered the possibility of splitting it into multiple modules, which you can develop and test separately? If each module can run as a separate process that would also have a performance benefit, because on a multi-core PC the load could be spread across multiple CPUs.

Richard.

Re: Internal Error 15
Post by JLThompson on Sep 9^th, 2014, 9:22pm

I can see how to split it into 2-3 separate modules running on 2-3 networked computers, with 30,000 lines each, or so. I seem to have 3,000 to 4,000 arrays, as an estimate. String variables? Hard to estimate. Same with handles, presuming you are referring to [placeMarkersOfThisSort]. If these three settings could all be increased to 5,000, though, it would likely encompass my program.

Re: Internal Error 15
Post by Richard Russell on Sep 9^th, 2014, 9:40pm

on Sep 9^th, 2014, 9:22pm, JLThompson wrote:

I can see how to split it into 2-3 separate modules running on 2-3 networked computers, with 30,000 lines each, or so.

With a careful choice of inter-process communication (IPC) you should be able to arrange that the modules can run either on separate computers or on separate cores/CPUs on the same PC.

Quote:

I seem to have 3,000 to 4,000 arrays, as an estimate.

It seems likely that this is the main limiting factor then. Incidentally LBB's arrays, unlike LB4's arrays, are not limited to two dimensions so I wonder if you can take advantage of that to reduce the total number:

Code:

    dim array$(3,4,5,6)
    array$(1,2,3,4) = "Hello world!"
    print array$(1,2,3,4)

Quote:

Same with handles, presuming you are referring to [placeMarkersOfThisSort].

No, they are labels, for which the current limit is 10,000. Handles are prefixed by # (e.g. #file, #dll or #window).

Richard.

Re: Internal Error 15
Post by JLThompson on Sep 9^th, 2014, 9:51pm

OK then, I would only need the array and string limits increased. I see what you mean about rewriting for different dimensions, but that would take far too much time at this stage.

Re: Internal Error 15
Post by Richard Russell on Sep 10^th, 2014, 11:11am

on Sep 9^th, 2014, 9:51pm, JLThompson wrote:

OK then, I would only need the array and string limits increased.

I have updated LBB to version 2.71 in which the maximum number of arrays and (non-local) string variables has been increased to 5000. There are no other changes.

Richard.

Re: Internal Error 15
Post by JLThompson on Sep 10^th, 2014, 7:00pm

Thank you very much, that is very generous of you. I tested it on my previous prototype, which has similar parameters, and it worked this time. Even found a number of bugs that Just Basic didn't catch. Hard to tell on a program with a variable run time, but it looks like it runs 2 to 6 times faster on this. This jumps my project ahead by months or even years, and saves me several hundred dollars to boot. Once again, thank you!