I feel 5.3.2 is the most stable release.
We just upgraded from 5.2.3 to 5.3.4 and have had no problems. But the site I am curently working at is very new to WebFOCUS, hence there are very few programs developed so far.
In the core product,there has ALWAYS been an 'EX', which established a 'new' environment. It was how MRE implemented it that was at fault. They (programming) did it similar to the -INCLUDE method, by putting the 'included' code in-stream. The reason for '-INCLUDE' was to both preserve the environment, and to support the inclusion of partial code (i.e. parts of a HEADING). Since we pointed out the error in their implementation, the EX method should be fixed soon, so you can invoke a new environment, AND pass variables to it, without the need for using GLOBAL variables.
francis, try this code on your 534
but remind me again what OS you're running on?
if its windows2003..have your server console open and be prepared to kill the agent. This fex triggers the infitite looping; you'll know because you'll see the echo pop out , but the little blue bar on the bottom is still running...
if so, as soon as you see the echo appear, kill your agent.
-SET &COUNT = 0 ;
-REPEAT LOOPEND FOR &COUNT FROM 1 TO 1000
-TYPE CYCLE NUMBER &COUNT...
TABLE FILE CAR
PRINT CAR MODEL SEATS
ON TABLE HOLD
This works for us on 5.3.2 (we're on Sun unix and I ran this in DevStudio, not MRE, which we don't have). The other errors that you mention sound like configuration problems.
CYCLE NUMBER 1000.
TABLE FILE CAR
PRINT CAR MODEL SEATS
ON TABLE HOLD
0 NUMBER OF RECORDS IN TABLE= 18 LINES= 18
0 STATISTICS OF LAST COMMAND
RECORDS = 18 SEGS DELTD = 0
LINES = 18 NOMATCH = 0
BASEIO = 6 DUPLICATES = 0
SORTIO = 0 FORMAT ERRORS = 0
SORT PAGES = 0 INVALID CONDTS = 0
READS = 0 OTHER REJECTS = 0
TRANSACTIONS = 0 CACHE READS = 0
ACCEPTED = 0 MERGES = 0
SEGS INPUT = 0 SORT STRINGS = 0
SEGS CHNGD = 0 INDEXIO = 0
INTERNAL MATRIX CREATED: YES AUTOINDEX USED: NO
SORT USED: FOCUS AUTOPATH USED: NO
AGGREGATION BY EXT.SORT: NO HOLD FROM EXTERNAL SORT: NO
thanks so much , suzy, for testing.
i may well go to 532 now as a last try.
I've got a real smart css-er working with me on this now. i'm installed on a brand new empty 2003 server..and css helped me install, so i'm thinking its not config, but more something whacky in the release... stay tuned...
Tried the looping test on WebFOCUS 5.3.4 AIX and it did NOT go into an infinite loop. This is as a self serve fex - we do not have MRE.
Tried the looping test on WebFOCUS 5.3.2 Windows XP and it did NOT go into an infinite loop. This is as a self serve fex and as an MRE fex.
this is looking more and more like a win2003 incomp., in my totally unschooled opinion.
does anyone have 534 MRE working, and can you open Report Assistant? and do you see the 'compute variable' tab anywhere? (My RA won't open so i can't check)
I was able to open Report Assistant in MRE 5.3.4..and to be honest have never used that before...used Report Painter or just code...did not see a Compute tab or anything that could be used for one. We are on Windows 2000.
i just talked to a css brainiac and we found it
its obscure, sort of cute, but we found it.
its in a wee dropdown box over the measures area next to the SUM or PRINT buttons.
on this looping issue, css wizards have determined it is a win2003 server incompat issue...due to IIS 6.
so...everyone...DONT PUT 53n on WIN2003...
but, even on win2k, 534 has a foc1517 error EVERYWHERE on everyfex...so they've gotta fix that in 535. so the only production choice is 532 with the hotfix for @echo looping...but 532 isn't readily available for download... css is arranging to make it avail on an ftp somewhere ...
and the WONT apply the hotfix to 532 for us...
and give us a full working version...I think they should...so we can get to work
[qb]update: july 22...css has determined that 533 and 4 are not compatible with iis6, which is whats in win2003 server. the wicked infinite looping, having nothing to do with &echo, is caused by iis6.
Do the ii6 problems extend beyond MR?
2. . . . -INCLUDE fex2 doesn't help if fex2 has statement labels that might conflict with existing labels inside fex1, or if fex2 needs to be run multiple times inside fex1... goto label conflicts.
According to Art's post, multiple includes of the same fex should not create label conflicts in 5.3.3. If fex1 and fex2 have conflicting labels that's a different story (but I don't see why that would have worked in 5.3.2)
Going back a few versions (it might have been called desktop at that time) I seem to remember a change to the way labels were actioned.
If I remember correctly (grey cells problem?) it was changed from the standard of picking up the first occurence of a label to picking up the last label of that name encountered.
I never could understand the reason for this, as I would always want to have unique labels even within parent fexes and their includes. Control the environment and not letting it control your approach.
That's when you have duplicate labels within a single fex.
In 5.3.2 and earlier, MR expanded fex2 at each occurence of -include fex2 in fex1, and handed the whole file to eda as adhocfex. In that case, adhocfex would contain fex2's labels multiple times (one per MR include of fex2). When it encountered the -GOTO XXX in the second -includes's code, it would goto the one it had already encountered, and infinite loops or other problems would ensue.
Since 5.3.3 makes an MR -INCLUDE into a server side -INCLUDE, there is no ambiguity: When EDA's FOCEXEC parser notes that label XXX was already encountered at line nnn of fex2, its conclusion that it can proceed directly there is correct.
|Powered by Social Strata||Page 1 2 3|