[Gambas-user] Debugging Gambas (again)

Cedron Dawg cedron at exede.net
Tue Apr 30 15:56:31 CEST 2019


Briefly, when I said "generalize the discussion" I also meant generalizing to any interpreter, not just Gambas.  This is pretty clear from the second line you quoted me on.

As for the points:

1) Gambas programs (except for the JIT component) aren't executed, they are interpreted, they are a data stream.  The actual execution (native machine code) is the interpreter running.  So it is the interpreter that is causing the seg faults based on the values in the byte code stream, or passing along bad values to system calls/external functions which then can seg fault.  A seg fault is initiated by the processor, which calls a kernel handler, which sends a signal, .....

Either the interpreter checks and catches bad values before it executes them and does a clean, hopefully informative, "error exit" or it doesn't check and seg faults.  You keep inferring fragile code is an error, I don't see it that way.  That is not the same as saying that Gambas should not have any fragile code, which seems to be such a strong sentiment for you that you don't make the distinction and say that is a bug.

2) "some sort of internal stack" means a program defined LIFO (Last In First Out) storage structure and is not referring to the processor stack (ESP/EBP register controlled on x86).  Whether it is overflow protected is implementation dependent.   Tobi's testing of GoSubs vs Calls with 10000 level deep recursive calls without overflow indicates to me that Gambas has overflow protection (and automatic expansion) built into its internal process control stacks.

Returning to the OP's particular situation.  I think it is safe to say that if the interpreter running the OP's Gambas code seg faults then there is either fragile interpreter code or a bug in the interpreter code.  If it is a bug, of course it should be fixed.  If it is a matter of fragility, whether to add further parameter validation code (i.e. look for errors) is a judgement call in which the answer is likely to be yes.

My initial assessment of this situation was that it was a case of interpreter fragility and invalid OP data.  Now, I'm not leaning at all.  Still curious though.


----- Original Message -----
From: "Jussi Lahtinen" <jussi.lahtinen at gmail.com>
To: "user" <user at lists.gambas-basic.org>
Sent: Monday, April 29, 2019 5:46:16 PM
Subject: Re: [Gambas-user] Debugging Gambas (again)

Did you even read what I wrote? You got nothing correct. 

Your points; 
1. In C, not in Gambas. We were talking about Gambas. 
2. Can happen in C, but the stack protection probably stops the execution before that happens... and again, not in Gambas. 
3. Same as point 1. 

"Suppose the interpreter provides a "SendToRemotePeer" ..." 

Wrong, does not happen in Gambas. Unless of course there is bug in Gambas. 

"Should an interpreter be coded to be robust? A general purpose usage one, like Gambas, yep. Should it be absolutely robust? I'm not sure that is even theoretical possible." 

I'm not talking about "absolute robustness", I'm talking about fixing bugs when we find them. Proper error message tells quite a much more than just sig11. 


More information about the User mailing list