| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 29.1 | | ORPHAN::BRETT | | Tue Sep 18 1984 18:26 | 4 |
|
Its ANNOUNCED - actually tomorrow, the 19th Sep '84.
/Bevin
|
| 29.2 | | EGEV00::SCHERRER | | Sat Sep 22 1984 16:54 | 10 |
| CONGRATULATIONS!
***************
The announcement arrives just on time... A Danish Company
was validating an ADA compiler for the VAX... I hope we
don't wait too long before we sell it. Is a 6 months field
test period really necessary for that compiler as it has
already passed the Validation Suite?
Patrick
|
| 29.3 | | LATOUR::AMARTIN | | Sat Sep 22 1984 23:45 | 10 |
| The Danish company sent me a brochure in the mail that arrived on Friday.
They must have copied my address out of the membership list published in
Ada Letters, because they didn't use the usual ACM mailing labels with
membership numbers on them.
I mailed photocopies of everything, including the envelope front with
"Par Avion" stamps to Bevin as soon as I opened it on Friday. I haven't
absorbed the contents of the brochure. If it has any comments about
"validating" or "validated", I'll post word.
/AHM
|
| 29.4 | | ORPHAN::BRETT | | Sun Sep 23 1984 00:14 | 27 |
|
We have known that the DDG compiler was coming since just after we certified.
Quality Unknown.
Our field test will NOT be 6 months - we go to FT on Nov 1 and SDC March or
April (forget exactly when). The reason for the big gap between validation and
FT are *HUGE* improvements in code quality and compiler speed. For instance
we expect to cut HOURS off our running time on the validation suite, and I
have made changes that reduced code from 400 bytes to 50 bytes in some common
cases, and from 10,000 bytes to 0 bytes for our STARLET support!
There are a variety of reasons a long FT is not thought necessary, including
the use of pre-FT sites other than the developers using the compiler.
We will have FT sites with 10's and even 100's of thousands of lines of Ada
code already developed (??? knows how), that will have never been run through
our compiler before so we hope to get the bug reports in pretty quick.
We will NOT be doing active development during FT.
Some time ago the decision was made to get to validation early, and not worry
about the gap between validation and frs. One reason for this is now our
compiler can be bid in Govt. contracts, but doesn't have to be avail. until
after the contract is awarded. Of course the announcement had to go with
validation since that was public knowlege.
/Bevin
|
| 29.5 | | LATOUR::AMARTIN | | Sun Sep 23 1984 13:11 | 44 |
| Was it a product announcement, or was it merely a press release that you
passed validation? I hadn't heard whether you got some reporters in the
ZK cafeteria, or did an announcement in some trade show, like Podunk DECUS.
From all the flamage surrounding Jupiter, I gather that we would be in
court right now if Jupiter had ever been formally announced before it
was cancelled. (I assume formal announcement means that DEC will sign
legal contracts to supply a product).
It would be an interesting difference if the product was announced only
because a validation certificate has been issued. In order to get
advance sales, yes. To get a good start in the market because of random
other companies coming out with Ada compilers, yes. But the way I have
seen validation of Fortran-10/20 handled, the fact that the public can
see through federal validation reports that DEC has validated a compiler
for X on system Y is not sufficient by itself to announce the product.
Fortran-10/20 comes up for re-validation in March every year. Both last year
and this year we were not even in field test for the versions that were
validated. We just validate the compiler when the old certificate times
out, and only announce it after it has had field test exposure, just like
we are supposed to. I don't recall the contingincy plans for the case
where we would have to cough up a validated compiler before announcement
because some customer threw a tantrum. I assume that they would either
get the save sets of the validated compiler, or would be licensed as
an alpha test site where we could keep an eye on them.
As far as field test goes, it seems necessary to have one and let some users
give the product a thorough going over. For one thing, the validation
suite probably doesn't test the performance of the product, and the
two projects that I can think of that were never announced (Jupiter and
Pascal-10) failed because of performance problems, not lack of
functionality. Also, I know that the Fortran validation suite doesn't
catch 2 of the 3 remaining unimplemented features in our field image
product. Ada has a much better validation suite, but it is a much
more complex language also. I'll be interested to hear if training
an Ada system to pass the suite really means that it is well trained
to pass actual applications.
Good luck; it sounds like things have been done well, and are going
well. I'm pretty sure you guys will succeed this time, but everyone
reading this note will be on some project that blows it in a big way
before they retire. It doesn't pay to count your chickens before
they are shipped to SDC.
/AHM
|
| 29.6 | | LATOUR::AMARTIN | | Mon Sep 24 1984 13:19 | 24 |
|
One of the enclosures in the Dansk Datamatik Center (DDC) Ada announcement
is a news release that says:
"
The portable Ada(R) Compiler developed by Dansk Datamatik Center (DDC) has
been certified by DDC on the U.S. DoD Ada Compiler Validation Capability
Test Suite currently in effect (release 1.4).
All disputes concerning the ACVC test programs have been resolved with
the DoD, and the validation runs have been scheduled.
"
Was Vax-11 Ada certified against ACVC 1.4? Since you already have your
testing runs done and you have a certificate, it sounds like you got the
jump on this group, at least.
Another enclosure says that the price is US$26,000, but there is an
introductory offer of US$19,000 for the period 1-Sep-84 to 15-Oct-84.
The price doesn't include a symbolic debugger which will is claimed to
be available Real Soon Now.
Can you quote any license prices for the DEC Ada system?
/AHM/THX
|
| 29.7 | | ORPHAN::BRETT | | Mon Sep 24 1984 18:25 | 20 |
|
Yes, we did announce a product including pricing.
Yes, we did validate against ACVC 1.4.
Pricing
730 and up, $24,900
uVAX $ 4,980
VAXcluster $14,940 - I presume this means per machine
Educational --I believe very heavy discounts, but don't have the
--figures.
I am amazed at how close we and DDG came with our pricing without cooperation!
One big advantage is it will probably force DG/Rolm to lower their current
price ($80,000) even before our FRS, thus cutting into their revenue, and
also impact their machine sales.
/Bevin
|
| 29.8 | | ORPHAN::BRETT | | Mon Sep 24 1984 18:26 | 6 |
|
Yes, this is MUCH more expensive than other VAX/VMS compilers - but very
reasonable according to non-DEC people at the Federal Computer Conference
where we announced.
/Bevin
|