[Implemented] Don't allow undeclared variables

Justin · Post by **Justin** » Tue Oct 28, 2003 12:08 am

it can produce hard to find bugs, just make a mistake in a variable name and i'll be assigned a 0 val producing unexpected results without errors

with declaring i just mean initializing, a=0 , a$="" ..

(Implemented as 'EnableExplicit')

stan · Post by **stan** » Tue Oct 28, 2003 2:44 pm

Hi,

I completely agree with you about the bug possibilities, on the other hand it would mean decalring all the variables in my "old" programs, so maybe a slightly better solution : a switch in the compiler options allowing to choose between being obliged to declare or not.

Bests.

Stan.

Psychophanta · Post by **Psychophanta** » Tue Oct 28, 2003 8:14 pm

But you are free to declare or not.

For my opinion that should be a serious restriction.

blueznl · Post by **blueznl** » Tue Oct 28, 2003 9:58 pm

a compiler option would indeed by a serious bug finding helper...

GedB · Post by **GedB** » Wed Oct 29, 2003 1:01 pm

Something like VBs Option Explicit would be a great help.

pthien · Post by **pthien** » Wed Oct 29, 2003 2:33 pm

Agreed. I've stumbled on this a couple of times already. Simply mistyping a variable name when testing for a condition can take several minutes to spot. Perhaps a warning rather than stopping compilation?

Thanks,
Phil

Jan Vooijs · Post by **Jan Vooijs** » Fri Oct 31, 2003 1:26 pm

@all, we agree but how about FRED, he has to oke this and built it in!!

Please FRED build this "option explicit" into Pure this is needed!!

Just simple an option into the sources or an extra parameter option!
I prefer the first IN the sources!!

Jan

freedimension · Post by **freedimension** » Fri Oct 31, 2003 1:56 pm

In the Meanwhile, for Global Variables you could just use a Structure. Structures are checked against wrong members.

Code: Select all

Structure global
  myVar1
  ...
EndStructure

global.global

global\mayVar1 ;this will not work

Jan Vooijs · Post by **Jan Vooijs** » Fri Oct 31, 2003 10:20 pm

@freedimension, thanks for the tip but that complicates the matter to a large extend!! That is NOT workable for me, any one??

Jan

Shopro · Post by **Shopro** » Mon Nov 03, 2003 4:39 pm

I agree to adding an option to force the declaration of variables.

tinman · Post by **tinman** » Mon Nov 03, 2003 10:07 pm

Someone once wrote a "source code checker" for this situation for the original Blitz Basic II on the Amiga. The source is available should anyone want to port it to PB (should be similar enough to not be a massive job - probably not a small one though :) and use it as a stop-gap until this gets put into PB.

Blitz-2000 site - although there seems to be problems with the server currently.

Blade · Post by **Blade** » Mon Nov 03, 2003 10:08 pm

Agree.
50% of possible common bugs will never appear again!

einander · Post by **einander** » Tue Nov 04, 2003 10:20 am

Simply asking "New Var : XXX?"
each time that a new variable is introduced would be a great help.

gomio · Post by **gomio** » Fri May 14, 2004 7:09 pm

You can try this:

(begin of program)
structure DefaultType: endstructure
deftype.DefaultType
(...)
foo = 0 ; <--- error
bar = ":-)" ; <--- error, etc.

NoahPhense · Post by **NoahPhense** » Fri May 14, 2004 7:15 pm

I like the 'Option Explicit' idea myself..

- np

PureBasic Forums - English

[Implemented] Don't allow undeclared variables

[Implemented] Don't allow undeclared variables

Re: Don't allow undeclared variables