Issue 192 revisited -- still broken in interactive mode.

[GoogleCodeExporter]

r605 was intended to address issue 192, but does not do so for interactive mode.


What steps will reproduce the problem?
1. Enter the following at the command line / within picoc interactive mode:
Marks-MacBook-Air:picoc-608 heilpern$ ./picoc -i
starting picoc v2.2 beta r608
picoc> #include <stdio.h>
picoc> 
     > void printArray(void) {
     > #define SIZE 10
     >     int array[SIZE] = {5, 4, 3, 9, 1, 8, 6, 7, 5, 2};
     >     printf("4: %d\n", array[4]);
     > }
picoc> printArray();
Segmentation fault: 11



What is the expected output? What do you see instead?
Expected output is "4: 1" rather than the crash/segfault.

I've verified that running the test case for this change does in fact work; the 
crash only occurs from interactive mode.



What version of the product are you using? On what operating system?
r608 / OS X Mavericks

Original issue reported on code.google.com by [email protected] on 3 Mar 2014 at 9:31

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

This crash happens at the same line of code as bug #190 and for the same 
reason: pc->InteractiveCurrentLine is null. It's another re-entrancy issue. 
This time, while executing printArray(), picoc is re-evaluating the #define. 
This, of course, is not how a compiler would handle a preprocessor statement. 

It's also bad style to put a #define inside a function, but it's legal C.

Original comment by [email protected] on 5 Aug 2015 at 2:42

[GoogleCodeExporter]

Although this crash is similar to bug #190, the fix that worked for 190 
(preserving the lexer state) needs to be applied at a different line of code. 
I'm not sure exactly where, but it's somewhere in the call stack for the crash.

Original comment by [email protected] on 17 Aug 2015 at 3:11