I know Report Caster has a different way to process Focus commands in a procedure than MRE.
I have a report running fine from MRE but From Report Caster, it's giving an error on a hold file created by a MORE command. Is there a problem using MORE when running thereport from Caster?? If yes, than I don't know how to change my code, I really need to use MORE...
There wqs a problem with MORE in Report Caster in version 4.36 - I don't know if it is fixed in later releases. What I used to do is hold the output of the MORE in a file, and TABLE FILE the hold file. (One extra unneccessary step, but it doesn't take long).
in 525, on win2k, no problem with MORE
i run them alot every day.
don't know about later versions or other opsys
Try doing a ? HOLD holdfilename after each of the hold files you are going to append together and see if the hold files have the same field formats. Sometimes MRE is not as picky as Castor.
Ok. I did a ? HOLD after each hold file and they are exactly the same. About holding the output of the MORE in a file, and TABLE FILE on it: that's already what I'm doing. Here's my code:
TABLE FILE BUDGEQUA
ON TABLE HOLD AS MONTDAT FORMAT ALPHA
DEFINE FILE MONTDAT
TIMESORT/A1 = 'Q';
TABLE FILE MONTDAT
ON TABLE HOLD AS BUDGETQU FORMAT FOCUS INDEX ZPROFCTR_KEY
Caster says: THE DESCRIPTION CANNOT BE FOUND FOR FILE NAMED: MONTDAT
It's working fine in MRE, what is going on???
We are using Web Focus 5.2.5 on a Win 2003 server
|<Pietro De Santis>|
This probably won't help, but stick a -RUN after each END statment.
1. Are you certain that it works in RCaster without MORE?
2. It is possible that the agent is creating the .mas file, but is parking it in a directory that it doesn't search when it encounters later references to it.
The destination for a hold file's .mas file is normally the current directory (the agent's temp dir), but that may be over-ridden by APP FOCHOLD (when app is enabeled), or by FILEDEF HOLDMAST (when disabled).
So, I'd add
at the top and
after every END or one-line command, to correlate the messages to the commands;
DIR (assuming it's Windows)
before the TABLE and after its END to see what's up.
but jack, the problem is in Caster...and you won't have such an echo from a caster job, is that right? Since it works ok running live,
Flip, can you just try one stupid thing..and change the filename from MONTDAT to some other goofy name...just to see if its a java reserved word we know nothing about.
And here's another thing to try...in your caster job, stick this command in after the hold
CMD COPY MONTDAT.MAS D:\IBI\APPS\TEST\MONTDAT.MAS
and see if caster's agent actually has that master and can actually write out somewhere.
(make some goofy test area to write in that is NOT in your path, either APP or EDA)
The filename wasn't the problem, I've tried using other names, but thanks for the idea, it's really when I'm using MORE to concatenate many files into one.
It's a big report and I was using MORE several times in my code, Hopefully I was able to replace them one by one by using focus formats and the USE command. After each change, I tested it and Caster bombed on the next hold file with a MORE phrase. Now they are all changed and the report is finally working!!!
But I'm really disapointed by Report Caster, I just don't understand why a report runs fine in MRE but errors out when scheduled it, It's really bad and I'm p... off I lost 2 days of work because of that. Next time, maybe I won't be able to replace my code with something else.
I will open a case.
|<Pietro De Santis>|
Take a look at this document:
Best Practices - Considerations When Using ReportCaster Version 5 Release 2
� The following features are not supported in scheduled procedures:
� PCSEND - WebFOCUS reporting feature to store report sections in different files. This feature is only supported for immediate requests using the WebFOCUS CGI or ISAPI. It is not supported with the WebFOCUS Servlet.
� Multiple TABLE or GRAPH requests with HTML format. Internally, these are multiple HTML documents. Only the result of the first TABLE or GRAPH request (without a HOLD or SAVE command) will be distributed.
� MORE - This command is used to concatenate data sources.
[qb] but jack, the problem is in Caster...and you won't have such an echo from a caster job, is that right? [/qb]
"In ReportCaster 5.3 the -SET &ECHO is supported and the echo'd code will display in the ReportCaster Log and in the HTML view source." (http://techsupport.informationbuilders.com/sps/92231022.html)
Another thought: MORE didn't complain, it's the subsequent TABLE that couldn't find the hold .mas file. Suppose RCaster behaves like some of its ancestors, and won't search (for a .mas) in the temp dir unless a HOLD was performed earlier in the session. And suppose it's fooled by the absence of the word "HOLD" in the last leg of the "TABLE ... MORE ... END" construct. Sooooo ... The old MVS prescription is to add
CHECK FILE CAR HOLD (or a TABLE FILE HOLD for one record from some other data source)
to the head of the fex. That might just clear it up. (Doubtful, since you say you've put HOLD FORMAT FOCUS before the MORE request. Still worth a stab.)
Thank you very much Pietro, I won't open a case since they already know it's an issue...
I wish I'd see that document before. At least I now know how to use "USE" to concatenate several focus format hold files.
Yeah, I just found the same caveat in Troubleshooting:
Product: WebFOCUS ReportCaster
Primary OS: Windows
Task error: UNRECOGNIZED COMMAND FILE CAR
Receiving the above error in the ReportCaster Log when scheduling a FOCEXEC
with the MORE command.
MORE is not supported in ReportCaster
(Not much of a "solution". )
So, why does my MORE work just fine? in all my caster'd ETL jobs, in 525. and has worked fine since 435. ????
I use isapi for my webserver calls from my UI, but my mre and Caster are stricly servlet.
[And as for multiple TABLE's in an email, piece of cake. That BP could use an update.]
[I had heard that caster echo was to be a 53 feature, but Flip is a 525 programmer...nice to hear it's a reality]
Let me go out on a limb:
Note that Flip's symptomatic msg is not "what does MORE mean", as in the doc I cited, but "where's the MFD (used in step 2, created by step 1)". I am also puzzled - doesn't RC (like MR) submit the requests to WFRS for execution, which is where the execution-time error msgs would eminate. It's the same reporting server as plain old webfocus, why would it suddenly have a senior moment w/r/t universal catenation, just because RC is the submitter?
-- Ergo, this error msg really comes from the pre-processor within RC, which fails to "notice" the HOLD appearing earlier in the TABLE request, and (Pharoah-like) cannot find this name in his book of MFDs. So maybe -mrnoedit is all that's needed.
Flip, run the job in mre and watch it in the agent..watch the files being made. Then do the same from Caster. Do it at some off hours when you can easily identify the agents.
Your agents are at
where nnn is the number corresponding to the Tscomid in your server console, where you normally watch jobs.
SEE if your masters are getting made, or just see what the difference really is.
|Powered by Social Strata|