Announcement

**Eric Pearson** · 12 Mar 2001, 05:14 PM

I won't try to address all of your items, but...

> The compiler doesn't check array dimensions.

That is definitely by design. It allows multi-dimensional arrays to be addressed "linearly" as well, which is a useful function. Especially considering that certain functions like ARRAY SCAN always treat arrays as having one dimension. (I understand that this will be documented better in the next version on the help file.)

> PB/CC Execute problem

IMO that would cause as many problems as it solves. Personally I would not like it if I had to manually close every program I run in the IDE, whether it is PB/CC or PB/DLL. The easy solution is to add a WAITKEY$ to the end of your program. Or SLEEP.

> Functions shouldn't need to be declared in order.
> I noticed that this isn't the case for Subs, so I
> figure it shouldn't technically be necessary for
> functions either.

The CALL keyword tells the compiler that a "word" is a sub, so it can continue. Without that it can't be sure whether a word is an undeclared variable, or a function that it hasn't encountered yet, or a simple typo. Which error message should it display?

-- Eric

------------------
Perfect Sync Development Tools
Perfect Sync Web Site
Contact Us: mailto:[email protected][email protected]</A>

**Bern Ertl** · 12 Mar 2001, 06:14 PM

Daniel,

I've gotten into the habit of adding leading spaces before every
use of a variable in my code. The Find feature works then, also
CTRL-Left & CTRL-Right when moving the cursor from word to word.

Code:

Call MySub(Var, MyVar)  --->  Call MySub( Var, MyVar)

If PB can add the Split Screen editing, I'll agree with you about
the auto-updating of the Sub/Fn box. Otherwise, I enjoy begin able
to CTRL-PgUp to edit equates/globals at the top of my file and then
use the Sub/Fn box to return to where I was.

I like the explicit (%) declaration of equates.

Otherwise, your list seems very reasonable to me.

------------------
Bernard Ertl

[This message has been edited by Bern Ertl (edited March 12, 2001).]

**Edwin Knoppert** · 12 Mar 2001, 10:21 PM

I really dislike ANY option... except Option explicit.
It's therable if you get an include file with option base 1 for example.
Never liked Defint or else.

in my vision they never should be invented in the first place.

------------------

**Tom Hanlin** · 13 Mar 2001, 01:50 PM

Quite the list! Thanks.

Tab setting doesn't work for all blocks.

Actually, it does, but the current auto-indent handling uses a somewhat idiosyncratic
approach with variable-sized indentations according to the length of the keyword used
to start the block. This is likely to change in the future.

Search engine problem.
If I have lines such as: "Call MySub(Var, MyVar)", there is no
way for me to successfully do a search for "Var" using PB's find
button. If I search for "Var" without "Whole Word" matching,
then it finds "MyVar" also, and if I do Whole Word matching, it
skips over it entirely.

The IDE expects a "whole word" to have white space on either side. This is also
likely to be subject to future adjustment.

Keyword mixed case doesn’t always work.

A few keywords aren't capitalized by the IDE. This will be fixed in the next
update.

if I have a global variable, and I declare a local variable of
the same name, it should warn me of the potential problem.

A common technique here is to use an appropriate variable naming scheme to help you
differentiate by scope. For example, prefix all global variables with "g_".

The Sub/function box at the top right of IDE allows you to jump
directly to a function. This is great. Now if I manually move
into a function, then that box should be updated to reflect the
current function that I'm in also.

This is already on the wish list. It appears that implementing this might impose
a significant performance penalty, however, so it may not be practical.

Is _ (underscore) a valid character for use in variable names?

An underscore is a valid character in any symbol name (although symbols must
begin with an ASCII letter). Yes, the requirements for symbol names should be
better documented.

Can UDT elements share the same name as a function or PB
keyword?

Function, yes. Keyword, no.

What does "Animate Delay" do?

It specifies the delay between steps when you use the Animate option in the
debugger.

Do numeric equates, which are preceded with % support only
16-bit integers?

Any integers (up through 64-bit QUAD).

Explain how AND and OR do bitwise and boolean double-duty

All such operators are inherently bitwise in BASIC. The exception is in the
use of AND in a PowerBASIC conditional statement, where evaluation is done
in a Boolean fashion by default (to allow for early-out optimization). Use
parentheses around the conditional expression to force bitwise evaluation
in all cases. Yes, this will be covered in more detail in future revisions
of the docs and help files.

------------------
Tom Hanlin
PowerBASIC Staff

**Daniel Corbier** · 14 Mar 2001, 11:36 AM

Tom, is the "early-out" optimization the same as "short-circuit"
evaluation (as explained in the MSDN), where the second argument
is not evaluated if the first argument of AND is false? You
mentioned the optimization for AND, but is it also done for OR?
For instance:

If A=1 Or SlowFunction(A) = 10 Then ...

Would it skip the second argument? If not, I'd like to suggest
that as a feature. Perhaps there might be some complications
which would make it difficult to implement. I noticed that in
the .NET version of VB, it seems like they will separate the
functions so that you have "AND" (boolean AND) and "BitAnd",
and "OR" separate from "BitOr". But if PB can do it with one
operator for bitwise and boolean, then it's even better.

As for the Whole Word search, I do hope that it is changed.
Bern suggested adding a space before variable names. But the
problem is extended to commas and quotes as well. For instance
consider:

Dim A As Double, B As Double, C As Double

Now I want to Search/Replace the word Double with Long. That
won't work either. You'd have to make it work around parenthesis
(opening and closing), brackets, quotes, commas, etc... Although
I haven't verified it just now, I'm pretty sure that the various
IDEs I've used, and Wordprecessor, Wordpad, Notepad, Browser type
programs, likely all do Whole Word searches the way I had in mind.

Eric, VB doesn't seem to mind the order of function definitions.
So I figure it must be possible with PB as well. Perhaps it
would require a second pass, and take a fraction of a second
longer to compile, which would not bother me. PB can include an
IDE option that you can toggle depending the programmer's
preference on the matter. I tend to have a preference for fewer
and shorter lines of code (w/o sacrificing legibility) to
accomplish the same task.

PB can have an IDE option for Waitkey$ for PB/CC. That way we
both would have it the way we want it. Or if PB implements the
Template idea, then one can add Waitkey$ to the start-up
template. The template would have the PBMain declaration, along
with whatever you usually want to start up with. When I want to
just print something out real quick, I find myself loading VB
to do it, because it takes fewer keystrokes. If PB's quick
load-up time was combined with an optional start-up template
that would be ideal.

Concerning array dimension checking, if it's by design then that's
fine (as long as it's documented). But I found some other little
things, which I'm not sure would be by design. Consider the
following code, for instance, which compiles without a problem:

Code:

$Compile Exe
Option Explicit

Function PbMain()
   Dim MyArray(10) As Long, MyArray(10, 10) As Long
   Dim OneDim(10) As Long, Answer(10) As Long
   Dim Answer As Long

   Answer = OneDim(1, 2, 3)

End Function

Notice the duplicate definition of MyArray; OneDim is
declared with one dimension, but used with 3, and Answer
is declared both as an array and as a regular variable.

------------------
Daniel Corbier
UCalc Fast Math Parser
http://www.ucalc.com

**Tom Hanlin** · 14 Mar 2001, 12:42 PM

"Early-out evaluation" is also referred to as "short-circuit evaluation", yes.
Also yes, this is automatically done for OR operations. No special handling is
needed in that case, because the "truth" of a bitwise OR is the same as the
truth of a Boolean OR. Such is not the case for ANDs, which is why PowerBASIC
optimizes further by assuming a default of Boolean ANDs in conditionals.

Yes, it would require at least one additional pass by the compiler to avoid
the need to declare functions if they are referenced before they are
encountered in the code. No, this is not the sort of modification that is
likely to be undertaken lightly.

There is no problem with DIM'ing Answer as both an array and a scalar
variable, as it will always be clear to both the compiler and the programmer
which one is referenced. I'm not sure why duplicate DIMs of MyArray go
through. I'll refer it to R&D.

------------------
Tom Hanlin
PowerBASIC Staff

**Mike Doty** · 15 Mar 2001, 01:50 PM

Code:

'#Register NONE 'required with FLUSH (hopefully reported)
DIM ALL
#COMPILE EXE
FUNCTION PBMAIN AS LONG
  DIM Count AS LONG
  OPEN "DATA" FOR BINARY SHARED AS 1
  MSGBOX "Count before flush" + STR$(Count&)
  FLUSH #1
  MSGBOX "Count after flush" + STR$(Count&)
END FUNCTION

------------------

**Lance Edmonds** · 15 Mar 2001, 03:52 PM

Thanks Mike, but that one has been known for some time, and it has been discussed on the BBS a few times too.

------------------
Lance
PowerBASIC Support
mailto:[email protected][email protected]</A>

**stephane fonteyne** · 17 Mar 2001, 03:13 AM

Wishlist for PBDOS, PBCC en PBDLL

Build-in or Adding : DEFAULT PARAMETERS and OVERLOAD ROUTINES

------------------

**Mike Doty** · 18 Mar 2001, 11:57 AM

Lance,
A post somewhere on known problems would be
useful. Newer users wouldn't know about flush.

------------------

**Lance Edmonds** · 18 Mar 2001, 02:21 PM

Understood, however, responsibility for such a posting is under control of R&D and PowerBASIC Management, not Tech Support.

To be clear, the problem with FLUSH is that it can corrupt a register variable, so using #REGISTER NONE just within the SUB/FUNCTION that FLUSH occurs solves the problem.

------------------
Lance
PowerBASIC Support
mailto:[email protected][email protected]</A>

Announcement

Wish list and bug report

Wish list and bug report

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment