home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!sun-barr!ames!pacbell.com!att!ucbvax!ROO.FIT.EDU!SAHARBAUGH
- From: SAHARBAUGH@ROO.FIT.EDU
- Newsgroups: comp.lang.ada
- Subject: MI, a hope for Automatic Test Equipment
- Message-ID: <9208251111.AA07899@ajpo.sei.cmu.edu>
- Date: 25 Aug 92 11:07:00 GMT
- Sender: daemon@ucbvax.BERKELEY.EDU
- Distribution: world
- Organization: The Internet
- Lines: 47
-
-
- Bob Crispen writes, in part:
- >email to me if you
- >think that there is a domain in which MI is both a natural part of a
- >domain-specific software architecture and a safe and effective >technique
- >(e.g., system integration or component debug took *less* time than
- >previously).
-
- I have hopes but nothing that matches the past tense "took"
- in your request. Here it is for comment anyway.
-
- I am doing work on the IEEE standard "Ada Based Environment
- for Test". It is definitely a DSSA, where the domain = Automatic
- Test Equipment (ATE).
- The hope is that test objects can be utilized to reduce the cost
- of producing Test Program Sets (TPS's). NCR reports a 5:1
- productivity in using an object-oriented approach. I don't
- know if it involved MI or not.
- I could imagine a scenario where I needed to create a TPS
- for a PC clone. I would derive my TPS from a test object
- class named PC_Test_Program_SET and turn off any methods
- that didn't apply to the particular PC I need to test (e.g.,
- turn off the tests for hard disks if my PC didn't
- have a hard disk).
- I further imagine that the PC_Test_Program_Set object
- would inherit the PC_motherboard_Test_Program_Set and I
- would have to communicate that I want to use the method
- for a 486DX cpu and the methods for other CPU's would
- not be utilized.
- I would further imagine that the PC_Motherboard_Test_Program_Set
- would utilize a 486DX_Test_Program_Set from Intel which would
- include methods for 30MHZ, 50Mhz etc. and that I would have
- to state my clock rate and that this would be communicated
- somehow up 2 levels to the Intel test object.
- This is all a dream so far BUT if it is off base, unachievable
- etc. please say so and why.
- No small consideration is that the test system must be
- implementable in Ada. I suspect there is no use to
- consider Ada83 but rather look forward to Ada9X objects.
- If Ada9X objects don't allow multiple inheritence then
- the above inheritance will have to be programmed somehow.
-
- I have sent this to Bob who may put it in perspective
- when he summarizes the replies.
-
- sam harbaugh saharbaugh@ROO.FIT.EDU
- -----------
-