Craziest Progressism?

[GregTomkins]

Incented by the 'for each with rowid' post, our group spent all day looking for weird Progress behaviours. It's a long list, but here's my favorite!

def var h_char as char init today.
message h_char.

Of course, this doesn't actually matter at all. It does make you wonder what lurks in the depths, though?

 

[tommartinson]

Whatever you do, don't ever do this!

def var x as int.
def var y as int init 999.
x = x + 1. + y.
disp x.

 

[GregTomkins]

That looks correct to me? The '+y' would be interpreted as a single expression consisting of a unary + operator operating on a variable.

The unary + operator doesn't actually do anything in Progress, but lots of languages have it: What does the unary plus operator do?

 

[davidvilla]

Incented by the 'for each with rowid' post, our group spent all day looking for weird Progress behaviours. It's a long list, but here's my favorite!

def var h_char as char init today.
message h_char.

Of course, this doesn't actually matter at all. It does make you wonder what lurks in the depths, though?

I think you can set anything to a CHAR variable without quotes and it stores it.

define variable mychar as char init 12445.
disp mychar

result: 12445

define variable mychar as char init 11/5/15.
disp mychar

result: 11/5/15

 

[kolonuk]

What about:

RUN program.p(INPUT v_Var1, OUTPUT v_Var2) (v_RandomVariable).

This compiles fine...

 

[kolonuk]

...and technically

DEF VAR v_Logi AS LOGI.
v_Logi = ?.

is a Progressism - what other languages have a three-state boolean?

 

[RealHeavyDude]

IMHO - that is not quite right. The third state theoretically means that it is undefined. Compared with the primitive boolean in Java one might argue that his could be right. Compared with a Boolean object one might argue that it is the same as the object reference being null.

private Boolean myBoolean = null;

But yes, practically you can work with the "3rd state" being null.

Heavy Regards, RealHeavyDude.

 

[GregTomkins]

RUN program.p(INPUT v_Var1,OUTPUT v_Var2)(v_RandomVariable).

IIRC, 'v_RandomVariable' (and any number of similar tokens following it) will get passed to program.p as a run-time argument, which the called program may choose to ignore. So I think this is a case of 'unusual, but as documented'.

This is a variation of include files, and harks back to a bygone era of compile-on-the-fly unholiness (who else remembers when functions and internal procedures were the hot new feature? v7 maybe? I hear that even DEF PARAM was a cool new thing at one point, but I'm not quite old enough to remember THAT).

 

[TomBascom]

Ok. I'm showing my age but:

User defined functions were version 8

Internal procedures were v7

DEFINE PARAMETER was v5

 

[jongpau]

...and technically

DEF VAR v_Logi AS LOGI.
v_Logi = ?.

is a Progressism - what other languages have a three-state boolean?

I like calling that "Yes/No/Maybe"

 

[jongpau]

Ok. I'm showing my age but:

User defined functions were version 8

Internal procedures were v7

DEFINE PARAMETER was v5

Not so much language related so a somewhat off-topic but I certainly remember the struggle with the r-code size limitations (what was it, 32kb max for a .r?). And then of course there was of the "fantastic" code editor prior to v6 where one wrong keystroke could overwrite your hard work with an empty file. Oh yes, we are getting old

 

[jongpau]

DISPLAY NO-LOCK - 1 = SHARE-LOCK.

and

DISPLAY SHARE-LOCK - EXCLUSIVE-LOCK + SHARE-LOCK = NO-LOCK.

 

[GregTomkins]

Oh man, this must be a sign of serious mental health issues on my part, that totally made my day !!! In a similar vein:

buffer some-table:handle:FIND-FIRST("",6207).

 

[KrisM]

Not so much language related so a somewhat off-topic but I certainly remember the struggle with the r-code size limitations (what was it, 32kb max for a .r?). ...

That was 63kb

 

[Stefan]

We had lots of fun with the 64k e-code limits in 9.1 and defined variables to use for often used constants.

define variable ltrue as logical no-undo initial true.
define variable lfalse as logical no-undo.
define variable i0 as integer no-undo.

 

[grinder]

message chr(0) = chr(32).

 

[jongpau]

That was 63kb

You are right and also showing your age

 

[jongpau]

message chr(0) = chr(32).

Now THAT is obscure (and somewhat disturbingly scary?)

 

[GregTomkins]

My documentation-inhaling coworker points out that it is, at least, documented (in the CHR function):

The CHR function returns a null string if the expression yields a value outside of the range 1 to 65534 or the expression yields a value in the range 256 to 65534 and the value does not correspond to a valid lead‑byte.

Kind of reminds me of those ISO-9000 "lifejackets made out of concrete" stories, though.

 

[andre42]

RUN program.p(INPUT v_Var1,OUTPUT v_Var2)(v_RandomVariable).

IIRC, 'v_RandomVariable' (and any number of similar tokens following it) will get passed to program.p as a run-time argument, which the called program may choose to ignore. So I think this is a case of 'unusual, but as documented'.

This is a variation of include files, and harks back to a bygone era of compile-on-the-fly unholiness (who else remembers when functions and internal procedures were the hot new feature? v7 maybe? I hear that even DEF PARAM was a cool new thing at one point, but I'm not quite old enough to remember THAT).

Indeed. If you run this in a production environment without source code you will get a runtime error (STOP error, probably) because program.p can't be compiled. One time such an error slipped by our QA. Apparently they also had source code in addition to .r-Code.

My documentation-inhaling coworker points out that it is, at least, documented (in the CHR function):

The CHR function returns a null string if the expression yields a value outside of the range 1 to 65534 or the expression yields a value in the range 256 to 65534 and the value does not correspond to a valid lead‑byte.

And of course you have to keep in mind that string that only differ by trailing spaces are considered equal.