Hi,
We are debating on where to develop and keep Fex and html launch pages. We need to use report caster schedule button in most of the reports(which is available only in MRE). Would you Please tell me which place is the most efficient ?

Thanks,
DD

This message has been edited. Last edited by: Kerry,

I would definitely develop in MRE. You mentioned the schedule capability which must be hand code in Java script if you do not use MRE. You will lose the ability to use deferred processing if you only code a URL in MRE to reference a html launch page on the server. Change management tools are provided for MRE. You could lose some security controls when the new MRE security mode is introduced in a upcoming release. In a split tier environment, i.e. WebFOCUS Client on one platform and the WebFOCUS Server on another you will have to share the apps folders which requires additional setup or manage two separate sets of ibi apps directory structures. Sharing across different type of platforms is sometimes impossible, introduces security considerations and possible performance degradation depending on the configuration. Then you have the upgrade considerations when new releases are introduced when sharing or maintaining multiple ibi\apps folders. The only cited gain is the focexec does not have to be sent from the client to the server. In a single tier environment, this will be done over memory and generally in gigabytes speeds in multi tier environments. The gain is negligible vs. losing deferred processing,the ability to easily code the ReportCaster button, change management features and possibly other capabilities introduced for MRE development in the future.

One thing to consider/remember... If you do any includes in MRE then it will look for those fex's on the server.

Which also means that if you were to just have a shell of all includes in MRE then you would have minimal code on the MRE server. Most changes could be done on the server itself.

Change control can be done without interaction on the server. MRE it's all manual. I'm all about automation.

If you company has a NAS then the dirs can also be shared between all the servers. Which if you have the application in a cluster on your app server you need to have some form of shared disk regardless.

Thanks Ron and Mike, you provided us very good information. We really appreciate it.

Mike: We use CVS for version control. Is there any way to integrate CVS with MRE or Server Fexs?

Thanks Alot,
DD

CVS from a unix side? We created a custom product delivery that in short... Extracted from CVS and put into a standard set of dirs. Then tar'd up the file. The tar file was then processed on individual servers with the appropriate actions. On a WebFOCUS server it would copy the fex/acx/mas/profiles out to the correct dir(s).

We didn't integrate it with MRE. I don't think MRE has a command line. Plus with MRE it has to not only copy files out but update an xml (html) file that tracks which files exist and meta data about each one.

From what I've heard, in version 7.7/8.0 what ever they are going to version it as, most of this might not even be needed. Sounds like it "might" be database driven. But thats the future.

Thanks Mike.

quote:

it "might" be database driven. But thats the future.

Does that mean that IBI is finally going to implement some of the other features of RAMIS that
did not make it into the original version of FOCUS?

quote:

If you do any includes in MRE then it will look for those fex's on the server

Not quite. By default, any -INCLUDEd fex in an MRE procedure is searched for within the same domain. If not found there, you will get a "procedure not found" error. The only way you can get an MRE procedure to find a -INCLUDEd fex on the server is to put a -MRNOEDIT before the -INCLUDE.

Darin Lee,
Thanks for the clarification. I'm not a focus developer. I do administration. But when we were testing the MRE implementation the consultant was telling us that includes would be read from the server. It seemed very odd.

Just for clarification:

All fexes that are called with -INCLUDE are passed on to the server and then read by the server. But the client (MRE) searchs them only in the current domain.

-MRNOEDIT tells the MRE that it should do nothing with it except removing the -MRNOEDIT.

That is my knowledge so far.
regards,
Markus

quote:

Just for clarification:

All fexes that are called with -INCLUDE are passed on to the server and then read by the server. But the client (MRE) searchs them only in the current domain.

-MRNOEDIT tells the MRE that it should do nothing with it except removing the -MRNOEDIT.

Not quite right. Like I said in my earlier post, any -INCLUDEs in an MRE fex MUST be first resolved before ANYTHING is passed on to the server. That means that the only place MRE knows to look for a -INCLUDEd procedure is in the same domain as the fex being run (or an alternate domain explicity specified.) That is also the purpose of the -MRNOEDIT. It tells MRE that it DOES NOT need to resolve the -INLCUDEd procedure before passing the parsed code back to the WF server.

Once that code gets to the WF server (minus the -MRNOEDIT) the WF server ONLY knows to search its server path to find any remaining -INCLUDEd procedures. It DOES NOT have any way of knowing how to get back to MRE for additional code.

I don't know if that's the same thing that Computix was trying to say, but the beginning statement "All fexes that are called with -INCLUDE..." did not seem correct.

It was trying to say the same thing.

quote:

Originally posted by Darin Lee:
Not quite right. Like I said in my earlier post, any -INCLUDEs in an MRE fex MUST be first resolved before ANYTHING is passed on to the server.

Just one note for the "resolved":
As far as I have noticed by now, MRE resolves -INCLUDEs in the way that it provides the Rep.Server access to the fexes somehow.
Because you can do that in MRE:

-SET &FEX = 'TEST';
-INCLUDE &FEX
-EXIT
-INCLUDE TEST

Without the last statement it will fail, obviously because it cannot find the TEST.fex.

Regards,
Markus

It isn't really providing the reporting server access to the fexes. MRE is just resolving all of the code to be -INCLUDEd before it talks to the reporting server. The reporting server cannot access anything throught the MRE environment. (Aside from a direct directory mapping which is not very common.)

That was what I've heard before as well. But it does not explain why the code I wrote before could possibly works.

An MRE-fex is not able to work with a statement like '-INCLUDE &FEX'. It will tell you, it cannot find the fex.
AFAIK the MRE-Preprocessing does not include resolution of DM-Variables.
Hence, MRE cannot know, which file to read and what code to insert instead of '-INCLUDE &FEX'.

Why then does the same code work, when you place the correct '-INCLUDE TEST' somewhere in the code?

Regards,
Markus