>>dubious validity. For example, you say "It seems to be easier to get(posted 8755 days ago)>>Microsoft FrontPage working well when the Web server it uploads files to is
>>running Microsoft Internet Information Sever rather than when it is running
>>open-source Apache." I don't think this is true.... >
> I had thought that the server extensions were the key: that they
> worked *well* on IIS and were very buggy elsewhere... > Well, that's an example of violation of standards -- it's like what MS did with Java. They did similar things with their browser; though of course so did Netscape. Some of these "extensions" are useful, but some are just deliberate incompatibilities inserted to make sure the competition's software will look as if it isn't working. Whether these things are bugs or features often depends on where you're standing....
>>Another shaky statement is the assertion that "Software for minicomputers >>stagnated in the 1980s because each brand's version of the Unix operating >>system was incompatible with the others." I've used many of those different >>"flavors" of Unix. While the differences are a nightmare from a system >>administrator's point of view, the differences are really quite minor from >>the user's perspective. > > But not from the viewpoint of the software producer... Phil > Greenspun, for example, strongly recommends Suns because Oracle > develops on Suns, and then tackles other dialects of Unix...
An interesting point. I think of myself as a software producer, though it's mostly for my own use. But I don't use database products like Oracle. But I've run on many different Unices from AT&T SysV to Berkeley, with Xenix and Linux somewhere in between. (Then there are the rather peculiar ones like IBM's AIX, which I've also had to work on.) For the most part, if you stay away from systems programming and stick to applications and shell scripts, it's easy to write fairly portable code for all these.
Oddly enough, it was Sun's software that gave me the problems that were cleared up by replacing it with open-source versions.
So I suppose it depends on what you're doing. I think of Sun as a producer of scientific workstations; it seems odd to me that a business-software outfit like Oracle would use them. My own work is all scientific programming.
I had an interesting year back in 1992 developing software for the European Southern Observatory's MIDAS data-analysis system. Their stuff has to run on a dozen or more different platforms; they had recently switched from VAX VMS to UNIX, so they even had to run (still) under VMS. They basically had their own operating-system layer that ran on top of the actual OS, so my problem at that time was in dealing with the quirks of various Fortran compilers. Lots of things that conformed to the ANSI Fortran standard did not actually work on one or another of the several systems, so I had to reduce all my code to a sort of lowest common denominator. Fortunately I had read a bit about how to write portable code (as well as having had to port my own for over 30 years) so I already was writing fairly portable stuff. It still surprised me how many deficient compilers there were out there. That experience, despite these difficulties, really impressed me with the value and importance of having standards -- especially clear, unambiguous ones.
-- Andy Young