nsb@THUMPER.BELLCORE.COM (Nathaniel Borenstein) (07/31/90)
Am I the only person who has ever wanted to compile Andrew on a Dec 3100 at a site where we have only a binary UNIX license? As far as I can tell, it is impossible to compile Andrew at such a site, because you don't have the -g0 versions of libc.a or libm.a, and you have no way to recreate them because you don't have source code! Can anyone provide any helpful hints as to how to go about making Andrew work on a binary distribution of the Dec 3100 version of Ultrix? Thanks. -- Nathaniel
zs01+@ANDREW.CMU.EDU (Zalman Stern) (08/02/90)
Nathaniel Borenstein <nsb@thumper.bellcore.com> writes: [...] > Can anyone provide any helpful hints as to how to go about making Andrew > work on a binary distribution of the Dec 3100 version of Ultrix? > Thanks. -- Nathaniel Call up your DEC sales office and politely explain that you are from Bellcore and that you would like to compile Andrew on your DECstation 3100. Be sure to make it clear to them that if Andrew does not work, Bellcore will never ever buy another piece of equipment from DEC. Seriously though, I hear that Ultrix 4.0 will ship with -G0 libraries. Perhaps things do get better every great once in a while... Sincerely, Zalman Stern | Internet: zs01+@andrew.cmu.edu | Usenet: I'm soooo confused... Information Technology Center, Carnegie Mellon, Pittsburgh, PA 15213-3890 *** Friends don't let friends program in C++ ***
sean@DSL.PITT.EDU (Sean McLinden) (08/04/90)
Re: compiling Andrew on a DECStation 3100 We have done it and it AIN'T fun. Basically, we had to get a source for the C library (we used a combination of System V and BSD sources), minus the assembly language coded stuff, compile it, and then load in things that were coded in assembler after resolving all external references. I wouldn't want to do it, again. You also have to get sources for libm and libn (or libln). To me there are two reasonable alternatives: 1. DEC should make a -G 0 version of the C library available on WRL or one of those machines as a gesture of goodwill. They should also do it because it doesn't make a whole lotta sense to have the -G option if you have to use a library that is compiled with some other value than the one that you want. In spite of numerous requests they have yet to do this which is a disappointment and, perhaps, why they are posting quarterly losses. 2. There is no good reason why the C library sources should not be distributed with each system. After all, everyone HAS to implement it, and it doesn't contain any information (or at least, it SHOULDN'T) that should be proprietary to a single implementation of C or Unix. Certainly the assembly language support would hardly disclose any sensitive information. This is especially important as compiler options become more sophisticated since the weak link is a static C library. It's always easy to say "I wish the GNU people would do this..." so I won't, but it would be nice if each vendor would contribute such to the GNU (or other) effort(s). I can't send you the library sources that we used because of the System V stuff (ironic, eh? you being at Bellcore!), but you are welcome to the binaries. Personally, I wish that the distribution of a C library for Andrew on the MIPS had been negotiated with DEC upfront. As I understand it, ITC has a copy, internally, but I'm not sure what, if any, are the restrictions on their redistribution. Sean
zs01+@ANDREW.CMU.EDU (Zalman Stern) (08/06/90)
sean@dsl.pitt.edu (Sean McLinden) writes: > [...] > Personally, I wish that the distribution of a C library for Andrew > on the MIPS had been negotiated with DEC upfront. As I understand > it, ITC has a copy, internally, but I'm not sure what, if any, > are the restrictions on their redistribution. > > Sean The ITC does not have a copy of this code, Academic Computing does. I suppose the license covers all of CMU but we (the ITC) did not compile the -G 0 libraries and have no interest in doing so. (I doubt any ITC employee has compiled ATK on a DECstation since late last year.) The ITC has already been extremely generous in "donating" my time to getting ATK running on the DECstation. When I originally volunteered to do the port I figured it would be easy (and I'd get a DECstation in my office). But noooooo... It turned out to be two orders of magnitude more difficult than I expected. Most of the difficulty was do the the -G 0 libraries, but MIPS' very own symbol table format didn't help either. (In all fairness the format is pretty good but figuring it out sans documentation was a pain in the ass.) This is an unofficial position of course. Tom can comment if he feels it necessary. Sincerely, Zalman Stern | Internet: zs01+@andrew.cmu.edu | Usenet: I'm soooo confused... Information Technology Center, Carnegie Mellon, Pittsburgh, PA 15213-3890 *** Friends don't let friends program in C++ ***
dhc@lanl.gov (Dave Carter) (08/13/90)
i'm having the same "woes" as others, it appears, trying to get andrew to compile on a 3100. it seems that some people have done it, however. i don't expect to be getting ultrix 4.0 for quite some time, though. could some kind soul make available to me (through ftp) the andrew package with the fixes you used to compile this on a 3100? i would appreciate it. thanks! - dave dhc@lanl.gov
dhc@lanl.gov (Dave Carter) (08/15/90)
okay, from the mail i've received, i'm convinced i need to get a copy of the lg0 libraries in order to build andrew on my 3100 (running ultrix 3.1.) unfortuneately, getting anything from our local dec people is like pulling teeth, only not as quick. anybody out there have these libraries where i can ftp them? or even better, has anyone yet fixed andrew so i can build it on my 3100? if so, i'd really like to ftp the kit, complete with patches. thanks a lot! - dave dhc@lanl.gov