Re: RSE performance against very large source files

Jon,

Um, it's not exactly all in one source file. There is one source library for each application system, a system being an application library and a data library. The source files in each source library are divided by type, i.e. QRPGLESRC, QDDSSRC, QCLSRC, QCMDSRC, QCBLLESRC, etc. You're suggesting dividing the source files up into smaller units. That might be a good idea, but what's my compelling business reason for doing that, considering the changes required to move all that source around, adjust Hawkeye and Implementer to deal with it, educate 150 developers on how to find the production source the new way, make sure it passes muster with the auditors (even though they have no reason to care), etc.? This is very much an 'If it ain't broke, don't fix it' environment, and the vast majority of my co-workers either don't think it's broke or don't think it's worth a lot of effort to fix it. It's not like I'm anyone important enough to change things by edict.

As someone who always worked in small shops before coming here, it's absolutely astounding how hard it is to change anything from the bottom in a large organization like this. After 4 years, I still shake my head regularly.

Dave Shaw
Mohawk Industries

----- Original Message -----
From: "Jon Paris" <Jon.Paris@xxxxxxxxxxxxxx>
To: <wdsci-l@xxxxxxxxxxxx>
Sent: Monday, February 04, 2008 2:07 PM
Subject: Re: [WDSCI-L] RSE performance against very large source files

On 4-Feb-08, at 1:38 PM, wdsci-l-request@xxxxxxxxxxxx wrote:

I find that my list of members in a source physical file growing
larger the
more I adopt modular practices. I would only expect them to grow
in number.
How would they be separated out differently? Logically separate
them into
differenlty named source physical files?

Well I got the impression that the OP had all source members for all
applications in a single file.

I agree that more modular practices increase the number of files, but
I tend to group relates subprocedures in a single source anyway.

Also I would associate a subsystem with a single file rather than a
whole system.

I should point out that I find large lists unmanageable in PDM as
well as in WDSC/RDi - so I'll use many techniques to avoid it.