I have read the MFS.PDF, read many posts on implementing an MFS and still am unclear on a couple things.

1) Logged into Sysprog

2) created a new stored procedure MY_MFS which has 1 line

call MY_MFS_CODE(args,…), compiled.

3) created new stored procedure MY_MFS_CODE, compiled.

4) EXEC run set_mfs "tablename","myapp",1

Do I have to do this in Sysprog. If so, I am concerned about OI upgrades?

Do I need to do a database save (OI v.3.7.5)?

Can I just edit the SYSTABLE record for tablename and manually

add the MY_MFS to ?

On the code end of it. If the code coming in to the mfs is a WRITE

what is the most efficient way to get the "before" look of the record?

TIA

Hmm why the two stage approach? We'd just write an MFS (say AUDIT.MFS) then from system editor

RUN SET_MFS "TABLENAME","AUDIT.MFS",3

If you don't do it from sysprog use

RUN SET_MFS "TABLENAME","AUDIT.MFS*APPNAME",3

Then either way save the database definition or it will "disappear" when you next log in.

Easiest way is to save off the various values then set the code to 4 and call the chain to do the read, grab the record then restore the values.

The Sprezzatura Group

World Leaders in all things RevSoft

Thanks Sprezz,

Does it matter if the name of the MFS is MY_MFS or MY.MFS ?

The two stage approach is because Fld 2 of table A (A)

is accumulated in Fld 1 of table C (C).

Fld 5 or table B (B) is accumulated in Fld 2 of table C (C).

Instead of trying to nail down the table name by the handle I

created a procedure A_MFS and B_MFS for the respective tables.

They call ABC_MFS with 1 extra arguement TABLEID.

Seemed to me that having the MFS call another routine would give me

a little flexability but more so right now to clear my brain as I need

to get this in place and came up with this for now. Figured I would

go back and change later.

As for responding to the question of the recordbefore and recordafter

say the write code I don't think I quite follow.

Thanks again.

pmfji

bc] Does it matter if the name of the MFS is MY_MFS or MY.MFS ?

A rose by any other name .. the . will work but _ is preferred

bc] The two stage approach is because Fld 2 of table A (A)

is accumulated in Fld 1 of table C (C).

Fld 5 or table B (B) is accumulated in Fld 2 of table C (C).

If it is working for you now .. cool BUT i would suggest in the future to simply use the single routine approach it is, to my way of thinking, simpler/cleaner .. but your mileage may very

By the way I would use expendable so the compiling then running will make sure you have the 'newest' obj. Then when you are really happy remove it to speed it back up.

dsig@teleport.com onmouseover=window.status=the new revelation technology .. a refreshing change;return(true)"

David Tod Sigafoos ~ SigSolutions

Phone: 503-639-4240

SaveCode=Code

SaveHandle=Handle

SaveName=Name

SaveRecord=Record

Code=1

Call @NextFs(Code, BFS, Handle, Name, FMC, Record, Status)

OldRecord=Record

Code=SaveCode

Handle=SaveHandle

Name=SaveName

Record=SaveRecord

The Sprezzatura Group

World Leaders in all things RevSoft

Dave,

Thanks, I have been using the expendable. That was part of the mfs.pdf.

Still haven't changed the 2 proc. version. Haven't figured out

in the code how to tell which file is being processed. Will keep pluggin away though.

BC

Hmmm,

So your saying do a recursive call depending on the 'Code' arg. value?

So assuming I have 1 Mfs (which I really do anyway right now)

if I only wanted to process logic if a write happened your saying

check code for 'write' save the values; call @my_mfs('read'…)

then do the compare?

Ok, I made 1 routine.

I run from EXEC: in system editor

RUN SET_MFS "RECEIPTS","CB_MFS*CREDMAST",3

I get a 'red' results msg, just FS.

If I go to SYSPROG and do

RUN SET_MFS "RECEIPTS","CB_MFS",3

I get FS401 ' MFSLOG table has not been created or attached.'

MFSLOG is a table I created in MYAPP. It is a table I created to log

issues, errors that the cb_mfs might have during processing.

What gives here?

Thanks for the responses/replies; esp. DSig for the time on the emails.

My struggle was due to not having the shells. I was close, I needed

to have code for the INSTALL (22) and the correct STATUS (1).

I was using case statements (which would seem to me quicker than on gosubs (any benchmarks anyone?) and just plain struggling without the shell.

It would seem to me that some new MFS documentation for OI would be good, as well as a separate 'Online discussion'. MFS coding is trickey, sensitive and powerful but could also be de-constructive.

Thanks.

View this thread on the forum...