|
View previous topic :: View next topic
|
| Author |
Message |
ppleva
New User
Joined: 10 Dec 2007
Posts: 7
Location: Czech Republic
|
|
|
|
Hi all!
PL/I allows using composite symbols (e.g. += or -=) within source, which is well described (Language Reference - Program Elements - Single-Byte Character Set - Composite symbols).
Such symbols aren't listed in Preprocessor Statements list (Language Reference - Preprocessor Facilities - Preprocessor Statements; especially under "%assignment Statement"), hence they should not be accepted by PL/I Preprocessor.
However Enerprise PL/I compilers (I'm sure about "IBM(R) Enterprise PL/I for z/OS V3.R2.M0" and newer) accept such code! I don't complain  , but the question is: Where is this behavior described/documented? (I've googled for hours and didn't find anything  .)
Code sample:
| Code: |
%CSPROC: PROCEDURE;
DCL I FIXED;
I = 0;
I += 1;
%END;
|
|
|
| Back to top |
|
 |
PeterHolland
Global Moderator
Joined: 27 Oct 2009
Posts: 2481
Location: Netherlands, Amstelveen
|
|
|
|
| Enterprise PL/I for z/OS SC27-1460 |
|
| Back to top |
|
|
ppleva
New User
Joined: 10 Dec 2007
Posts: 7
Location: Czech Republic
|
|
|
|
Peter, thanks for response. Of course I've read Enterprise PL/I for z/OS Language Reference, even the latest one (V3R9), that's where I've started before I post to this forum 
However I don't see the answer there (in my original post I wrote where I was looking for it). Could you please be so kind and direct me to the particular chapter/page? |
|
| Back to top |
|
|
Robert Sample
Global Moderator
Joined: 06 Jun 2008
Posts: 8697
Location: Dubuque, Iowa, USA
|
|
|
|
| I clicked on the manuals link at the top of the page, pulled up the PL/I Language Reference manual, clicked on the Search icon, typed in += and did the search, and found chapter 9.2.2 Compound Assignment Statement as the second found page. This took less than two minutes, total. Perhaps instead of Googling, you should just go to the manual and use that as your starting point? |
|
| Back to top |
|
|
PeterHolland
Global Moderator
Joined: 27 Oct 2009
Posts: 2481
Location: Netherlands, Amstelveen
|
|
| Back to top |
|
|
ppleva
New User
Joined: 10 Dec 2007
Posts: 7
Location: Czech Republic
|
|
|
|
I've read the manual. I know what are composite symbols, I know how compound assignment statements look like. I know that "i+=1;" is correct PL/I statement, but I'm asking about PL/I Preprocessor statements (look at the code sample in my original post)
In another words:
A. I've verified that IBM Enterprise PL/I for z/OS V3.R2.M0 (and newer) compiler accepts compound assignment statement within preprocessor procedure.
B. In SC27-1460-09, Chapter 21. "Preprocessor Facilities", section "Preprocessor Statements" there is paragraph "%assignment Statement". In there I see a sentence: Compound and multiple assignments are not allowed.
I see a contradiction between statements A and B. Is this documented somewhere? |
|
| Back to top |
|
|
PeterHolland
Global Moderator
Joined: 27 Oct 2009
Posts: 2481
Location: Netherlands, Amstelveen
|
|
|
|
What the <hell> do you want to know?
Its there for you to use.
Am im getting psychic too? |
|
| Back to top |
|
|
Robert Sample
Global Moderator
Joined: 06 Jun 2008
Posts: 8697
Location: Dubuque, Iowa, USA
|
|
|
|
You don't understand the manual, for starters. The %assignment specifically states compound and multiple assignments are not allowed. However, if you read the syntax diagram on that very same page, you will note that it is referring to preprocessor variables, and only preprocessor variables. What is a preprocessor variable? From section 21.3 of the manual,
| Quote: |
| A preprocessor variable is specified in a %DECLARE statement with either the FIXED or the CHARACTER attribute. |
The code you posted in the original post has no preprocessor variables, so the %assignment restriction does not apply. You may place compound or multiple assignment statements that do not use the preprocessor facilities anywhere you want, including in preprocessor code. |
|
| Back to top |
|
|
enrico-sorichetti
Superior Member
Joined: 14 Mar 2007
Posts: 10873
Location: italy
|
|
|
|
if You are not satisfied or You find that there are <errors> with/in the IBM documentation
use the Readers' Comments at the end of the manual to send comments/feedback by snail mail
or search the IBM site on how to submit the same by electronic means
while the use of compound assignment statements in real code might, repeat MIGHT, yield some optimization advantages
( hard to believe...
it would be wise to look at the generated assembler code to check
I checked for C and there is no difference in the generated code ... i+=1; vs. i=i+1; )
<frankly> I do not see any advantage in using it in the preprocessor stage for preprocessor entities
( it adds useless clutter/complexity to the the preprocessor scanner/parser,
and the whole thing is being interpreted anyways )
but on the other side I do not see any issue in using it in the <generated> statements
maybe that' s what You see
and that' s what happened in the snippet You posted...
the three statements are just copied as is
they are not preprocessor's statements
anyway looks like that some of us have on the subject the same feelings as Rhett Butler |
|
| Back to top |
|
|
ppleva
New User
Joined: 10 Dec 2007
Posts: 7
Location: Czech Republic
|
|
|
|
Robert Sample & enrico-sorichetti, thanks for very detailed answer, I appreciate it.
I was confused by manual section 21.6 saying:
| Quote: |
| Preprocessor statements in a preprocessor procedure do not begin with a percent symbol. |
Hence I thought:  "DCL" (without %) is a preprocessor statement because it appears in a preprocesor procedure Then variable "I" in my sample is preprocessor variable If "I" is preprocessor variable it is not allowed in compound assignment statement.
However if "I += 1;" in my sample code is NOT preprocessor statement, then everything is clear. (Still I thought it is because (according to my weak PL/I knowledge) it seems to be evaluated in preprocessor step - see listing further.)
I admit my example with "I" might not be quite obvious... So what do you say about this example with "J"? Why this (line 18.1) works? (I'd expect something like "IEL0080I S 18 INVALID TEXT IGNORED FROM 'J' TO SEMICOLON." which I get with older compiler.)
| Code: |
15655-H31 IBM(R) Enterprise PL/I for z/OS V3.R2.M0 (Built:20031007)
- MACRO Source
0 Line.File
0 2.1 SAMP: PROC OPTIONS(MAIN);
3.1
4.1 %DCL J FIXED; /* preprocessor variable */
5.1 %J = 0;
6.1 %CSPROC: PROCEDURE RETURNS (CHAR);
7.1 DCL I FIXED; /* preprocessor variable too (?) */
8.1 I = 0;
9.1 I += 1;
10.1 IF I = 1 THEN RETURN ('PUT SKIP LIST (''I=1'');');
11.1 ELSE RETURN ('PUT SKIP LIST (''I<>1'');');
12.1 %END;
13.1
14.1 %ACTIVATE CSPROC;
15.1 CSPROC
16.1 %DEACTIVATE CSPROC;
17.1
18.1 %J += 5;
19.1 IF J = 5 THEN PUT SKIP LIST ('J = 5');
20.1 END SAMP;
15655-H31 IBM(R) Enterprise PL/I for z/OS SAMP: PROC OPTIONS(MAIN)
0 Compiler Source
0 Line.File
0 2.1 SAMP: PROC OPTIONS(MAIN);
3.1
15.1 PUT SKIP LIST ('I=1');
15.1
19.1 IF 5 = 5 THEN PUT SKIP LIST ('J = 5');
20.1 END SAMP;
15655-H31 IBM(R) Enterprise PL/I for z/OS SAMP: PROC OPTIONS(MAIN)
- Compiler Messages
0 Message Line.File Message Description
0 IBM1114I W 19.1 Comparands are both constant.
IBM1043I I 19.1 SYSPRINT is contextually declared as FILE.
- Component Return Code Messages (Total/Suppressed) Time
0 MACRO 0 0 / 0 0 secs
Compiler 4 2 / 0 1 secs
0 End of compilation of SAMP
|
|
|
| Back to top |
|
|
ppleva
New User
Joined: 10 Dec 2007
Posts: 7
Location: Czech Republic
|
|
|
|
| PeterHolland wrote: |
| What the <hell> do you want to know? |
I'm asking why PL/I compiler accepts something what is forbidden in PL/I language reference. And I'm asking where this should be documented. (I was thinking about this being a new feature which they forgot to mention in manual?)
| PeterHolland wrote: |
| Am im getting psychic too? |
I don't know, what do you think?  |
|
| Back to top |
|
|
prino
Senior Member
Joined: 07 Feb 2009
Posts: 1306
Location: Vilnius, Lithuania
|
|
|
|
| ppleva wrote: |
| PeterHolland wrote: |
| What the <hell> do you want to know? |
I'm asking why PL/I compiler accepts something what is forbidden in PL/I language reference. And I'm asking where this should be documented. (I was thinking about this being a new feature which they forgot to mention in manual?)
| PeterHolland wrote: |
| Am im getting psychic too? |
I don't know, what do you think? |
Please put this on the PL/I mailing list, which is read by IBM's head of PL/I development, Peter Elderon. He is the only authoritative source with regards to IBM PL/I standards.
Discussing this matter here is just a waste of time.
And if you want to go completely ape-shit the fact that PL/I has OO extensions is also not mentioned in the manual... And older versions of PL/I already had undocumented builtin functions. |
|
| Back to top |
|
|
Robert Sample
Global Moderator
Joined: 06 Jun 2008
Posts: 8697
Location: Dubuque, Iowa, USA
|
|
|
|
| Unlike your original post, at this point you have definitely identified what appears to be a discrepancy between compiler behavior and the documentation (there still could be an interpretation problem but if so then IBM needs to address that so either way there's an issue for IBM). The only source for an answer will be someone at IBM. |
|
| Back to top |
|
|
ppleva
New User
Joined: 10 Dec 2007
Posts: 7
Location: Czech Republic
|
|
|
|
| Thanks for answers. I think this topic can be closed now. |
|
| Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006
Posts: 19244
Location: Inside the Matrix
|
|
|
|
We would be interested to hear what you learn from IBM
d |
|
| Back to top |
|
|
|
|
