bob@imspw6.UUCP (Bob Burch) (01/07/89)
From Ted Holden of HTE.... ........................................................... From: Tom Almy Tektronix, Inc., Beaverton, OR. >Well, when my copy of C++ came, the BUNCH.EXE program, which is supposed >to make it posible to use Codeview (except for local variables, of course), >was a *ZERO* length file. Two phone calls later, they promised to send >out a disk with the program. A disk arrived with a label which had penned >on it "Bunch.exe", but the disk contained only two mysterious .lib files. >My conclusion is that BUNCH.EXE does not work. I haven't tried bunch.exe, but it definitely is more than a null file on version 1.07, the baseline. >My opinion is that the only way they will become "serious" and a "major >player" is via the steam roller effect C++ is having. I found about a >dozen bugs, worse than any compiler I have used. When I called with a >bug report I found that they were distributing *simultaneously* two >different versions (perhaps they will go with the one that produces the >fewest bug reports?). One of the compiler's most common error message was >to crash. >The package also shows immaturity in that there are no C++ example programs >on disk, and the only provided class is "stream", which is not documented. From: alann lopes >Things were going well until I got down to compiling >the test driver for my Robin-Hood hashing ADT. > >What I have is a common object and all the user records are derived from >this common object. >The problem arises when I try to compile lines where I'm either doing a >insert or retrieve. The compiler complains that it can't uniquely identify >the object (or something to that effect). >Note that this code compiled and ran fine on the VAX. >I called Zortech and they said that they could not help without seeing >the code, so I posted the source code on their Bulletin Board. >After waiting for a few days I called them and they still had not looked >at it, but they said that they would call me in a few days. After a few >weeks of no word I called them back and there answer was, "YEPP IT'S a BUG". >It appears that this is a serious, very serious BUG. >Is this supposedly complete C++ compiler unable to handle one of the >fundamental aspects of the language? >Has anyone had any similar problems with inheritance? >I would appreciate some feedback. Yes. However, they vanished with Version 1.07. > > > a VERY disappointed Zortech C++ owner > I've seen several such flames concerning the Zortech compiler. Basically, the people writing these flames don't seem to have any real idea of what is involved in writing a C++ compiler. From what I hear, the front end to the Zortech compiler represents about five times the coding effort of a typical C front end. It seems little wonder then that Walter Bright, by all accounts one of the brighter lights on the current software scene, has taken about a year getting the Zortech compiler from the first release version 1.0 to the present baseline version 1.07. They (Zortech) apparently went through several versions in quick succession getting rid of the last two or three remaining bugs. 1.06, in particular, lasted about a week. Zortech seems to be strong on talent (Mr. Bright et. al.) and on good-spiritedness (the idea of selling something like the C++ compiler for $100 or thereabouts) and a trifle short on organization. America is a free country, however; if you prefer dealing with a company which is longer on organization and shorter on talent, there are all kinds of options open to you. For instance, you could mail Bill Gates a check for $1500 and ask him to put you on the waiting list for a copy of HIS native code C++ compiler for MS-DOS. Don't, however, hold your breath whilst waiting for it; it's liable to be quite a little wait. Ted Holden HT Enterprises ...................................................................... Voice from the hereafter: "What could I have ever done for those @$%^*@'s to name a government programming language after me???....."
toma@tekgvs.GVS.TEK.COM (Tom Almy) (01/10/89)
In article <212@imspw6.UUCP> bob@imspw6.UUCP (Bob Burch) writes: >I've seen several such flames concerning the Zortech compiler. Basically, >the people writing these flames don't seem to have any real idea of what is >involved in writing a C++ compiler. From what I hear, the front end to the >Zortech compiler represents about five times the coding effort of a typical >C front end. Perhaps I don't know how much effort is involved in writing a C++ compiler, but I can recognize schlocky distribution practices and poor documentation. Just because a compiler is complex doesn't excuse: 1) Distributing more than one version at a time (I received my version 1.05 in Nlate October). 2) Distributing a compiler so buggy that it crashes even on small modules, billing it as other than a "beta" version. There is no way I could recommend this thing for other than toy use. 3) Inability to get one of the distribution files -- zero length on the distribution media, and junk sent in the mail after two phone calls. 4) No notification of a fixed version (1.07, what makes this "baseline"?), even after I called them (at my expense) giving them a detailed bug report in November. >They (Zortech) apparently went through >several versions in quick succession getting rid of the last two or >three remaining bugs. 1.06, in particular, lasted about a week. Fine, where is my version 1.07? I will phone them today and ask. >Zortech seems to be strong on talent (Mr. Bright et. al.) and on >good-spiritedness (the idea of selling something like the C++ compiler >for $100 or thereabouts) and a trifle short on organization. America is >a free country, however; if you prefer dealing with a company which is >longer on organization and shorter on talent, there are all kinds of >options open to you. For instance, you could mail Bill Gates a check >for $1500 and ask him to put you on the waiting list for a copy of HIS >native code C++ compiler for MS-DOS. Don't, however, hold your breath >whilst waiting for it; it's liable to be quite a little wait. Is $100 for a compiler that doesn't work a good buy? Even if written by a talented person? Does anybody know when Borland is coming out with Turbo-C++ at $150? Even their much-maligned (and deservedly so) Turbo-Prolhad fewer bugs than Zortech C++, and Borland send a free upgrade without even being asked! Tom Almy toma@tekgvs.TEK.COM Standard Disclaimers Apply