larry@JPL-VLSI.ARPA (06/25/86)
One group I'm helping has an Alsys Ada compiler for the AT and needs to import math and other routines. Has anyone solved this problem already? If so, we'd appreciate whatever help you can give (even copies of sample code without explanation would be helpful). ========================================================================== Several people have (rightly) taken me to task for suggesting a requirement for the Space Station SW Development Environment of "an operating system easily updated to the CAIS." Of course, that has to be restated in more specific terms, and in such a way that compliance can be measured. However, I doubt if any input to the Space Station management would have any effect. They seem to be firmly committed to IBM mainframes running MVS. It's what they've used with much success in the past, and people in the space program are (contrary to popular expectation) extremely conservative. Further, there are some very rational reasons for wanting IBM hardware and MVS operating system. Most of their people are expert in their use and won't require expensive, time-consuming re-training with all the attendant chances for catastrophe before becoming expert; a whole suite of SW-engineering tools are available from the shuttle program; and they don't have to worry about the vendor going out of business. Actually, I suspect the probability of many people ever using the CAIS is pretty small. The CAIS has a competitor that's already won the de facto status of an industry-wide standard, being used by 70-80% of all programmers in the U.S. And its supported by a vendor which stands to lose billions of dollars each year if an operating system standard is chosen that makes it easy to port SW to other vendors hardware. No, it isn't Unix and AT&T (which many supporters of the CAIS seem to consider their bete noire, judging from all the argument against it I hear). Its MVS and IBM. Larry @ jpl-vlsi.ARPA
alden@spp1.UUCP (06/25/86)
Larry, The arguement that "everyone knows MVS and IBM" so lets stay in the middle ages is most likely not going to wash as a reason to get out of the now Ada and soon CAIS requirement of DoD contracts. Time will tell. Most programmers do not interface to a significant extent with the underlying O.S. but rather use tool sets which differ from O.S. to O.S.: e.g. on Unix the popular editor is VI, on VMS its ED, on CPM its Wordstar, etc. Nothing precludes these tools being written in Ada and ported to the CAIS if the cost of retraining a given staff is higher than rehosting these tools. Thus from an O.S. point of view, I find the arguement of training cost not valid. The CAIS is to be used with Ada programs and the harder of the two to learn is Ada, not the CAIS. In addition, how much of the programming population writes tools? Small portion. Most of the difficult parts of the CAIS lie in those routines that will support tool writers in writing tools and not for those writing code to be run on target machines. The average programmer is not going to have to learn all that much to survive and produce good reliable code. If your complaint is that Ada can't be learned, that is another arguement. Personally, it's been my experience that the transition is not as painful as those who have yet to make the transition complain about. Further, for those who have worked on an Ada project, its been my experience that they prefer the language over others. Finally, no one has ever been able to show me statistics for a medium to large scale project where the total time (design, cost, test, and maintenance) was more using Ada VS other languages (Fortran, etc.) where the time was due to having to learn Ada. If Everyone agrees that there is a "software crisis" and that the cost of developing software is too high, then these same people must not argue for doing nothing - which is what not having to learn something new implies. ...Tony Alden TRW The above stated views are my own and do not reflect those of TRW or any management connected to TRW or any person living or dead connected to TRW.