megas@GTEWD.AF.MIL ("08374_MIKE MEGAS") (05/27/89)
Does anyone know what penalties are incurred for using an ACS sub-library
structure that's 8 or more levels deep? DEC's "ACS HELP SUBLIBRARY"
info is pretty explicit that there's a "practical limit" of 3 or 4
levels deep, but it doesn't say what that means -- slower compile time,
perhaps? If so, how much slower?
I'm looking at a very large CSCI. Rather than putting hundreds (or
maybe thousands) of code units into a single ACS library, intuition
tells me to break them up by category (according to which ones WITH
which other ones) and place them into separate ACS's that are
sublibraries of each other. Some folks tell me this is the wrong reason
to go to separate ACS libraries; but I'm just going by my experience
with VMS directories. It doesn't make sense to have dozens of
programmers working in the same VMS directory, so why should it make
sense to put the results of their work all into a single library?
Any comments?
The ACS library structure that I tentatively have looks like the
following. (This is for the integration "plane". Each of these
libraries will have its own sublibrary (not shown) for doing development
work.
--------
| | These top four c o u l d
-------- be combined into one, if the
| "cost" of the added depth
-------- outweighs the advantage of having
| | the code units distributed across
-------- these four somewhat-more-manageable-
| size ACS libraries.
--------
| |
--------
|
--------
| |
--------
|
------------------------------------------
| | | |
------- -------- -------- --------- The same logic
| | | | | | | | applies to each
------- -------- -------- --------- of these bottom
| | | | strings of three
------- -------- -------- --------- ACS libraries.
| | | | | | | | I would then have
------- -------- -------- --------- an ACS sublibrary
| | | | structure of depth
------- -------- -------- --------- 2 instead of 7.
| | | | | | | |
------- -------- -------- ---------