Proposed change to QB64 regarding CONST

[bplus]

Steve, do you make 1 BI for the 3 combined libraries or attempt to use the 3 BI's (separate and independent) already ready?

I think you attempt the latter choice and I think the former choice would solve the issue.

It's clear to me, both Fellippe and STxAxTIC are intuitively against the idea though The Disaster Example has not yet been presented.

So the conservative move is probably to wait ie for The Disaster Example to raise it's ugly head.

Constants don't use Type so that can't be a problem either unlike SHARED.

[RhoSigma]

If this is mainly about al that CONST TRUE.... thing, then i'm glad to use literal numbers in my libraries. Typing 0 or -1 is shorter than typing TRUE or FALSE.
However, the approch seems ok to me, if the CONST's name AND value are equal, then just ignore the double definition. But maybe it should be restricted to equal literal values only, not if the CONST's value is based on an expression or function result?

[TempodiBasic]

Hi

question: Do we teach to parser to ignore duplication of statement about CONST if they are made the same definition (for name and value)?
summarizing the different opinions showed until now:

let's make Pro and Cons
Pro
- no headache for error that are no error (a CONST doesn't change)
- high feasibility to combine library of different authors and/or ispirations

Cons
- no documentated  feature of BASIC language
- if we change one definition of CONST we fall in the issue of duplicate of name because the others are not already changed so we must search the others to change

other options...
--  make a strong changement of QB64 using

$IF TRUESET = FALSE THEN
    CONST TRUE = -1, FALSE = 0
    $LET TRUESET = TRUE
$END IF

and making a not backward compatibility with previous code.

This can be a feature of compiler activable by settings in menu (this can solve the backward compatibility issue)

-- create a .BI with only UNIVERSAL CONST from TRUE, FALSE... to OPENED, CLOSED, LOCKED, IS_DIGIT, IS_STRING  as a global constant file as heat of each program coded in QB64. Just like Steve's .BI for COLOR costants.

-- we can teach to parser/compiler to jump to translate (or to REM) the duplicates of CONSTs, so if coder likes  he doesn't care of duplication of the CONST otherwise he go down to delete duplications


Just a my real dubt
Does C++ compiler accept duplicate of CONST?

Thanks to read

[freetrav]

I'm going to suggest a "meet me halfway" sort of response, splitting the difference between those who say "Yes, ignore a redefined CONST if the value is the same", and "No, throw an error on a redefined CONST no matter what": If the value is the same, throw a WARNING, but let code generation proceed normally. If possible, when throwing the warning, identify the file and line number where it was first defined.

[FellippeHeitor]

Sorry we didn’t proceed with discussion here. We did come to that solution in the chatroom in the end.

[STxAxTIC]

Chatroom is #qb64 on irc.freenode.net

Come one, come all, stay the rest of your days!