Re: Program design to minimize record locks

Ok. I tried that. The program won't compile because of the following error:

*RNF3804 000002 Name FIELD1 in external description PMYFILE is not renamed; external name is ignored.
This appears not for every field in the data structure, but seems to be related to fields of the same name on the screen format.

There is no prefix keyword.

Pete



Booth Martin wrote:

But its not FILE01, Pete. :)
The file name used in the extname() word doesn't even need to be mentioned on an F-spec. The compiler will search the Library List and use the first file it finds that matches the extname() keyword.

So, if I understand things correctly (which is NEVER guaranteed!) we could do this:

F spec:
FPMYFILE UF A E K DISK RENAME(FMYFILE:FILE01)

D Spec:
DFileNameDS E ds extname(PMYFILE)


The key here is that there is no direct relationship within the compiler between the extname() keyword and the like-named file in the F-spec.
Are you by any chance using a prefix keyword on the F-spec? If so, then the data structure will be empty.

Pete Helgren wrote:

If I am not careful, I'll start chasing my tail.....

After your original suggestion Booth, I had the following:

F spec:
FPMYFILE UF A E K DISK RENAME(FMYFILE:FILE01)

D Spec:
DFileNameDS E ds extname(FILE01)

And remember my original concern that the RENAME in the F spec might bugger things but was needed since the /Copy "generic" I/O piece used FILE01 as the reference so in order for this model/template program to work, I needed to leave the name FILE01 alone in the /Copy member.

That wouldn't compile and Scott suggested that it wouldn't work because the file name FILE01 really isn't an external physical file it is a "placeholder" name used in the /Copy member. Scott suggested that I use LikeRec which I DID use and it would compile. However, the data structure FileNameDS contains no data after a read on the physical file as I expected. So I may be stuck here.

Extname won't compile. Likerec compiles but I end up with no data in the data structure.

Ideas?

Pete


Booth Martin wrote:

hmmmm.... I wonder if changing the LIKEREC back to EXTNAME might work? You do need the "e" - I'd forgotten that in the example.

Pete Helgren wrote:

OK! So the compile works but the I am not getting what I expected (maybe this IS harder than I thought).

Based on the prior posts from Booth, I thought I understood that the D spec for the data structure would hold a copy of the data read from the file "automagically". He said "The FileNameDS does in fact always have the most recently read data values, whether from the original read(n), the screen exfmt, or the locking read." So either I misunderstood or I am doing something wrong (that wouldn't surprise me).

Booth suggested the following:

D FileNameDS ds extname(Filename)
D OrigData s like(FileNameDS)
D NewData s like(FileNameDS)



This wouldn't compile until I changed the extname to likerec (and I had to add an "e" for externally described). I assumed that a read on Filename would automatically populate FileNameDS but that isn't the case. A read on Filename puts no data into FileNameDS. So I misunderstood or did it wrong.

Pete


Pete Helgren wrote:

Got it! I appreciate the clarification. It compiles now.
Thanks,

Pete


Scott Klement wrote:

Pete Helgren wrote:

A little wrinkle here:
The compiler seems unhappy with the fact that I have the RENAME

File spec has:

FPMYFILE UF A E K DISK RENAME(FMYFILE:FILE01)
DFileNameDS E ds extname(FILE01)

Did you mean to use LIKEREC?

A data structure defined with EXTNAME has absolutely no relationship with the F-spec. Instead, you tell EXTNAME the external name of a file, and it looks for it on disk to get the fields. Since the argument for EXTNAME is a FILE NAME (not a record format name) and it's looking directly at the file (not at the F-spec) I wonder if you really meant to use LIKEREC?

LIKEREC defines a structure to be like a record format that you've defined on an F-spec. So you could say

DFileNameDs ds likerec(FILE01:*input)

And that would work, because it's argument is the RECORD FORMAT NAME (which is what you renamed) and it's looking at your F-spec rather than directly looking at the file.