Re: IDE/Rational Apex experiences

[nospam@somewhere.ia.us (Robert S. White)]

In article <647cv8$1lf@dfw-ixnews3.ix.netcom.com>, bklungle@ix.netcom.com 
says...
>
>To answer your questions as best I can:
>1. From the time you press the enter key to a query to the return of the
>prompt.

  Query, as in a CMVC query to see version history or revision notes for
a given object?  I have never had them take longer than 30 sec max (and 
that is when the local Apex storage is very busy - like when an analyst
is running a signal processing simulation).  I should have mentioned that
our main data (source/Diana/library) storage for Apex is local to each
department - that is running of a RAID tower hooked to the Sun Ultra
that supports 6 simultaneous users.

>2. Apex is running on an SGI/Indigo2 with 128MB ram

  Is the data storage local or over the network from a distant file
server?

>3. All users are using SGI/Indigo2(s)

  Seems like that should be enough CPU power and ram for one user at
least.  

>4. Don't know about the closet but will check if you really think it means
>anything.

  Not much if the user's are executing on the machine right in front
of them.  It is not likely that ethernet switching will help a 
bottleneck with distant file server storage.

>5. They are using all kinds of IDEs also (including MSVC++5.0).
>6. The network is 10MB TP Ethernet.

  Ours is also and in my team only one or two users are sitting
right in front of a Unix workstation.  The other four or five 
simultaneous users are accessing our team's two workstations via
X terminal software running on PC's.  Page up/down is instant.
Hitting "Visit" on a module name referenced in source opens up a 
new window within a second (two max).  Syntax check is normally 
under a second.  Semantic check and Analysis check maybe 10 
seconds for a typical size module.  We do check out the Apex
executable code along with a license from a distant fileserver.
That does take a while (like the 20 seconds you talk about) for 
the GUI IDE to fire up.  We pay this penalty when we use the
Apex CMVC for engineering document version control.  But 
specification revisions are not quite as frequent as code module
revisions :)

  Come to think about it running a single Apex command from a 
c shell script takes the amount of time that you are talking
about - is that what you are doing?  Starting up Apex, doing one
thing (check out, check in, etc.), and then shutting down Apex.
That is what you do when you use Apex _only_ for version control
(only a replacement for SCCS?).  Do you stay in the Apex GUI (maybe
with the included choice of the GNU EMACS editor if you prefer it)?

  In general my team's (and every other team's that I have heard from)
has have a favorable experience with using Apex.  Seems like a very
good fit for the environment for which it is intended.  I have a good
experience with it just today in having it help me develop just
the right type of rep spec record structure using just data types and
attributes and +-1 (in Ada 83).  

  I have heard that the big benefit of using it with C++ is with its 
round-trip engineering with the layered Rose CASE tool product.  
Analyze and design in UML, generate code headers/templates in Rose.
Flesh out code in Apex, debug it and make it work.  Back to Rose and
do the graphical assisted comparison with the "as built view" to the 
"as modeled view".  Now that Apex has been Ada 95 validated with its
version 2.4 we are waiting for Rational to do the same thing with
Rose/Ada95. Some of us are lazy and want the computer to take care
of the mundane picky code generation details and just let us 
concentrate on the Big Picture (and fun/challenging algorithms).  Like
when you turn up the grammer checker for your word processor on high 
after you have done some stream of conscious composing.

_____________________________________________________________________
Robert S. White         -- An embedded systems software engineer
e-mail reply to reverse of: ia us lib cedar-rapids crpl shift2 whiter