gcw@lanl-a.UUCP (10/10/84)
This article is to warn others of the problems encountered with a Callan Unistar 200 running UN*X System V. We purchased a Unistar 200 from BMDP StatCat. The documentation was incredibly poor, with numerous serious mistakes. In addition: 1) the a, i, and c commands in ex cause a core dump; 2) uucp does not talk to 4.2bsd machines; 3) no means to generate nroff printer driving tables is provided; 4) the f77 compiler (from Absoft) is totally unreliable. Specific problems with FORTRAN are: 1) bad code is generated; 2) numerous deviations from the ANSI standard, usually significant; 3) does not use the floating point processor as documented. In response to numerous phone calls, plus now nasty letters, we get the run around (that's a System V problem, etc. etc.). Besides all this, the hard disk (43M) has begun to suffer from bad blocks. Let potential buyers beware. Gary C. White (lanl!gcw, csu-cs!csufwb!gcw)
jeff@ism780.UUCP (10/17/84)
3) no means to generate nroff printer driving tables is provided; The problem here is probably that the stuff to generate the printer driving tables might be considered source material. We argued about this among ourselves for a while here at Interactive before shipping IS/3. I don't remember whether the decision was that it could be included with a binary license or not.
ron@callan.UUCP (Ron Dunant) (10/18/84)
[] We at Callan are proud of the products we ship. We have well over 1000 units in the field to date, and yes there are problems. We try our best to address customer issues that are brought to our attention. I by no means think that support should be handled on the net in this manner, but feel the need to set Mr. White's one sided *story* straight. I suggest that in the future Mr. White contact Callan Data System's customer support group in a professional and business-like manner representative of the professional organization he is a part of. If Mr. White would have only listened to our answers, his article and my response would have been unnecessary. Mr. White's article appears here in its entirety. >> This article is to warn others of the problems encountered with a Callan >> Unistar 200 running UN*X System V. We purchased a Unistar 200 from BMDP >> StatCat. The documentation was incredibly poor, with numerous serious >> mistakes. First I would like to point out that Mr. White purchased a Unistar 200 with a preliminary version of UNIX (not UN*X) System V. Updates for preliminary software started to ship in mid June; I might add the update consisted of minor software changes. Why Mr. White has not as yet received his update has been taken up with BMDP. We have completed a revision of the documentation for the Unistar 200 and are preparing to ship updates of same. It should be noted that with all of Mr. White's talk about 'incredibly poor documentation' that he pointed out only four inconsistencies. We rely on customer feedback to help improve our products and documentation. We found two way communication very difficult with Mr. White. >> In addition: >> 1) the a, i, and c commands in ex cause a core dump; Yes, we knew that. Even before contact with Mr. White the bug was reported to Unisoft. In the most recent release sent to us by Unisoft, the ex bugs are fixed. The version of vi we received was broken along the way, but ex does work (vi and ex are linked). >> 2) uucp does not talk to 4.2bsd machines; This is correct. We recreated the problem and reported the bug to Unisoft. Excluding 4.2bsd machines, our implementation of uucp works without problems. Unisoft has notified us that they have found and fixed the problem and will supply an update this week. >> 3) no means to generate nroff printer driving tables is provided; AT&T has never written any documentation on nroff printer driver tables that I know of. There has been conversation regarding this on the net, but we have no immediate way of helping Mr. White. Judging from the fact that Mr. White posted his *article* about Callan, he must *know* how to use the net so we suggest that he explore net.wanted. >> 4) the f77 compiler (from Absoft) is totally unreliable. >> >> Specific problems with FORTRAN are: >> 1) bad code is generated; >> 2) numerous deviations from the ANSI standard, usually significant; >> 3) does not use the floating point processor as documented. Mr. White sent five short FORTRAN programs to our customer support group. Two problems were due to Mr. White not reading his manual; one was a 'feature'; and two were bugs which I might add are fixed in the current FORTRAN release. As far as Mr. White's accusation that the compiler produces 'bad code' (depending on one's definition of bad code), he has yet to supply us with any evidence of bad code. In reality, the code generated by the compiler is NOT too bad. As for the 'numerous deviations from the ANSI standard', we never advertised that our FORTRAN was ANSI compatible, it is an ANSI implementation. If Mr. White would have read his FORTRAN manual more carefully, he would have noticed Appendix J [CLEARLY marked] DEVIATIONS FROM ANSI FORTRAN 77. This appendix supplies a list of unimplemented features (total of 6), restrictions (total of 7), and enhancements. I might add that the unimplemented features and limitations are relatively minor. There were problems with Mr. White's FORTRAN release when used with the floating point processor. BMDP should not have sold Mr. White a floating point processor to use with his FORTRAN software, as we knew that it did not work. As a matter of fact, BMDP is no longer a Callan distributor. >> In response to numerous phone calls, plus now nasty letters, we get the >> run around (that's a System V problem, etc. etc.). >> >> Besides all this, the hard disk (43M) has begun to suffer from bad blocks. >> >> Let potential buyers beware. >> Gary C. White (lanl!gcw, csu-cs!csufwb!gcw) Considering that Mr. White's contact with Callan Data Systems and its associates is exemplified by his article posted on the net, I wish to rest Callan's case regarding attempted dealings with him. To answer Mr. White's last complaint in the article regarding bad blocks, which I might add he has not previously communicated to anyone at Callan: I suggest that he try the 'markbad' program. Every now and then a disk can 'suffer from bad blocks', and the markbad program will map the bad blocks to a known good area. If Mr. White's bad block problem continues, we can easily repair his hard disk. In closing, I would like to point out that anyone with a Unisoft port will have most of the same problems that Mr. White has experienced. As for the FORTRAN compiler, had Mr. White purchased our software maintenance agreement as we suggested he do, he would have been entitled to software and documentation updates at no charge. For Mr. White that would have meant the latest Absoft release with bug fixes (no known problems with floating point processor support) AND Absoft's 100% ANSI FORTRAN 77 (when released). Anyone who has had any experience at all with UNIX realizes that problems such as those brought up by Mr. White are not uncommon, nor can they be solved over night. Ron Dunant @ Callan Data Systems Engineering 2645 Townsgate Road Westlake Village, CA 91361 UNIX is a trademark of AT&T Unistar is a registered trademark of Callan Data Systems