Calculating PI with Nilakantha formula. *Bug*

[Ronald-Herrera]

A blue screen occur with the attachment sent.

crash-20250629-q1ukok84.log

[Marrin]

I can confirm this issue.

The error is dereferencing a string at address &H1697 which isn't there anymore. That address is very close to the end of the program (&H1653) and there is a string at &HFDCB that's not bound to a variable name, so I guess what happened was that string is the first argument of MID$(), then the creation of the string involved in the third arguments expression triggered a garbage collection that moved the string data of the first argument from &H1697 to &HFDCB but the code still tries to access it at the old address.

That seems to be a bug in PC-BASIC because the programs runs without those problems in GW-BASIC.

[Ronald-Herrera]

Thank you! Happy to help and test. Ron.

…

________________________________ From: Marc Rintsch ***@***.***> Sent: Saturday, July 19, 2025 8:27 AM To: robhagemans/pcbasic ***@***.***> Cc: Ronald-Herrera ***@***.***>; Author ***@***.***> Subject: Re: [robhagemans/pcbasic] Calculating PI with Nilakantha formula. *Bug* (Issue #251) [https://avatars.githubusercontent.com/u/2564163?s=20&v=4]Marrin left a comment (robhagemans/pcbasic#251)<#251 (comment)> I can confirm this issue. The error is dereferencing a string at address &H1697 which isn't there anymore. That address is very close to the end of the program (&H1653) and there is a string at &HFDCB that's not bound to a variable name, so I guess what happened was that string is the first argument of MID$(), then the creation of the string involved in the third arguments expression triggered a garbage collection that moved the string data of the first argument from &H1697 to &HFDCB but the code still tries to access it at the old address. That seems to be a bug in PC-BASIC because the programs runs without those problems in GW-BASIC. — Reply to this email directly, view it on GitHub<#251 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/BUDDMALLOS6EQAFI7KA32C33JI2TDAVCNFSM6AAAAACAMTDJ7WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTAOJSGMZDKMJTGQ>. You are receiving this because you authored the thread.Message ID: ***@***.***>

[robhagemans]

Reproduced locally. This may be difficult to root-cause and fix, since it only seems to after many iterations, i.e. a large number of string operations have happened. As @Marrin points out, garbage collection is a likely culprit, as it is something that also only happens after many strings have been created and deleted. Garbage collection should map all string pointers to their new positions but something failed here. Will need investigation.