spaf@gatech.UUCP (02/23/87)
If and when the network load gets to be too much (probably by the end
of July at the current rate), the first groups I'll cut here are
binary-only, or groups that carry mostly binary postings of programs.
There are a few reasons for that:
1) Binary isn't retargetable. A binary-only file is usable only on the
machine it was compiled for. With source, at least, you can make
(usually minor) changes and recompile for a different machine. No such
luck with binary.
2) Binary isn't adaptable. If a program is posted which does something
useful, but I want to "tweek" it for my environment, I need the
source.
3) Binary isn't fixable. If a bug crops up and a fix is made, the
entire binary file has to be reposted. A source code patch can be
posted in much less space and with much less net-wide impact.
4) Binary isn't verifiable. With all the crazies on the net, you
expect me to take some binary file posted to the net and install it on
*my* machine? Hah! I'm crazy, but I'm not *that* crazy!
There are also the usual problems with encoding method and size of
binary postings.
If you're going to start a group, I'd want source only. If you create
both source and binary, it is probable that sometime before too long
we'd only carry the source group.
--
Gene Spafford
Software Engineering Research Center (SERC), Georgia Tech, Atlanta GA 30332
CSNet: Spaf @ GATech ARPA: Spaf@gatech.EDU
uucp: ...!{akgua,decvax,hplabs,ihnp4,linus,seismo,ulysses}!gatech!spafK538915@CZHRZU1A.BITNET.UUCP (02/25/87)
As a major distributer of a binary only program (BTW I have put source for other programs on the net (did anybody ever use the screensavers I posted)) I have some comments to one of the postings about the a binary/source newsgroup (beware we are NOT connected to UUCP so we don't pay the bills): Gene Spafford spaf@gatech.UUCP writes >1) Binary isn't retargetable. A bina.................................. >2) Binary isn't adaptable. If a ...................................... Right, but utilities of general interest (in C and with Unixlike system calls) belong on Mod.sources or Comp.sources.unix (in source!), Comp.sources.atari.st or what ever it will be called should be for Atari specific code. Adaptions of existing code (if in C) should be done where possible with ifdef's. Don't forget that a lot of stuff written for the ST is not in C, an example (the one I know best:-)): UniTerm Source + Utilities 400kB Pascal source, about 200kB very machine specific 100kB Small assembler routines (about 80kB fonts) 15kB Fonts in GEM format 10kB Font to assembler dumper ? Font editor >3) Binary isn't fixable. If a ....................................... True, but a longer discussion about this was on Comp.os.minix, there are good reasons not to distribute sources, particulary if you offer some kind of support. >4) Binary isn't verifiable. With all ................................ That's one of the things a moderator is for. Simon Poole K538915@CZHRZU1A.BITNET
silvert@dalcs.UUCP (03/03/87)
In article <12621@gatech.EDU>, spaf@gatech.UUCP writes: > With source, at least, you can make > (usually minor) changes and recompile for a different machine. No such > luck with binary. > > If you're going to start a group, I'd want source only. If you create > both source and binary, it is probable that sometime before too long > we'd only carry the source group. > > Gene Spafford Gene raises an interesting point about the portability of source code. As Eunuchs (or whatever we call ourselves) we are used to source code in C which we know how to port between V at BSD. Non-Unix machines use other languages. CP/M and MS-DOS were long dominated by assembler and Turbo Pascal, and of course many micros use(d) BASIC. But the ST and other 68K machines use a variety of languages, and I don't think that UseNet supports much general source in these languages. I don't see any source in the modula-2 or f77 groups I subscribe to, and net.sources and mod.sources are automatically C. If there were groups called {net,mod}.sources.{C,mod2,pascal,asm,c++,...} then I would look for portable source code. Otherwise, I think that Gene has raised a red herring (which is properly the responsibility of older red herrings). Looking at this constructively, the basic question is whether source code is machine-specific or language-specific. If the latter, then Gene has a point, and we should set up language source groups. If the former, then the rest of us are on the right track. While I would like to see more machine-independent source code, it may be tricky -- what about all the GEM calls for the ST, for example. Is there even a portable way to search a directory in any language? -- Bill Silvert Marine Ecology Laboratory, Dartmouth, NS, Canada CDN or BITNET: silvert@cs.dal.cdn -- UUCP: ..!{seismo|utai}!dalcs!silvert ARPA: silvert%dalcs.uucp@seismo.CSS.GOV -- CSNET: silvert%cs.dal.cdn@ubc.csnet
ljdickey@water.UUCP (03/06/87)
In article <2425@dalcs.UUCP>, silvert@dalcs.UUCP (Bill Silvert) writes: > In article <12621@gatech.EDU>, spaf@gatech.UUCP writes: > > With source, at least, you can make changes and recompile ... > > If you're going to start a group, I'd want source only. > > ... the basic question is whether source > code is machine-specific or language-specific. If the latter, then Gene > has a point, and we should set up language source groups. If the > former, then the rest of us are on the right track. I agree with Bill on this one. I appreciate receiving his Modula-2 binaries. I do not have a Modula-2 compiler, and probably will not be getting one in the near future. I would also like to be able to exchange APL workspaces, which are in fact source code for an interpreter, but which to a non-APLer look like a "binary" because they have a lot of 8 bit characters in there. Even if I use the WorkSpace Interchange Standard (WSIS), which is a machine independent representation, it looks like a "binary" for the same reason. -- L. J. Dickey, Faculty of Mathematics, University of Waterloo. ljdickey@water.UUCP ljdickey%water@waterloo.CSNET ljdickey%water%waterloo.csnet@csnet-relay.ARPA ljdickey@water.BITNET UUCP: ...!watmath!water!ljdickey
apteryx@ucbvax.UUCP (03/12/87)
In article <12621@gatech.EDU> you write: >If and when the network load gets to be too much (probably by the end >of July at the current rate), the first groups I'll cut here are >binary-only, or groups that carry mostly binary postings of programs. There are a few reasons for that: ... >1) Binary isn't retargetable... >2) Binary isn't adaptable... >3) Binary isn't fixable... Neither is source, if you don't have a compiler for that language. I bet there are a bunch of Pascal programmers who don't use C, and vice versa. I don't have either language (I have Forthmacs), and quite a lot of my useful software has been binaries from the net. >4) Binary isn't verifiable. With all the crazies on the net, you >expect me to take some binary file posted to the net and install it on >*my* machine? Hah! I'm crazy, but I'm not *that* crazy! >Gene Spafford (:-) Anyone use peripherals containing dynamite? Or try new programs while especially valuable non-backed-up files are in the disk drive? Brian Peterson
jtr485@umich.UUCP (03/17/87)
In article <17795@ucbvax.BERKELEY.EDU>, apteryx@ucbvax.UUCP writes: > >1) Binary isn't retargetable... > >2) Binary isn't adaptable... > >3) Binary isn't fixable... > Neither is source, if you don't have a compiler for that language. > I bet there are a bunch of Pascal programmers who don't use C, and > vice versa. I don't have either language (I have Forthmacs), and > quite a lot of my useful software has been binaries from the net. > Brian Peterson But source in any language you know can be translated to another language you know. It just makes the retargeting and adapting more work. Have you ever disassembled a binary and recoded it? --j.a.tainter
sl@van-bc.UUCP (03/30/87)
In article <90@umich.UUCP> jtr485@umich.UUCP (Johnathan Tainter) writes: >In article <17795@ucbvax.BERKELEY.EDU>, apteryx@ucbvax.UUCP writes: >> >1) Binary isn't retargetable... >> >2) Binary isn't adaptable... >> >3) Binary isn't fixable... >But source in any language you know can be translated to another language >you know. It just makes the retargeting and adapting more work. > >Have you ever disassembled a binary and recoded it? Well, yes. There are some execellent dis-assemblers around these days. A case in point on the Macintosh is MacNosy. It makes mince meat out of most object programs. To the point that you can definitely re-target (at least to other 680X0's), adapt and fix. Within reason of course. Interestly, when doing this type of thing working with compiler generated code tends to be much easier to work with, except where a very good optimizing compiler was used. In my case I used MacNosy extensively to nose around in the Mac operating system, ROM etc, to find out how to actually get in and do things that where not documented all that well at the time. -- Stuart Lynne ihnp4!alberta!ubc-vision!van-bc!sl Vancouver,BC,604-937-7532
campbell@maynard.UUCP (03/31/87)
In article <90@umich.UUCP> jtr485@umich.UUCP (Johnathan Tainter) writes: >In article <17795@ucbvax.BERKELEY.EDU>, apteryx@ucbvax.UUCP writes: >> >1) Binary isn't retargetable... >> >2) Binary isn't adaptable... >> >3) Binary isn't fixable... ... >Have you ever disassembled a binary and recoded it? Well, yes, actually. The boot loader for my machine. And /etc/init. But just because it's possible to work this way doesn't mean I prefer operating without sources! I prefer a sources-only policy. Binaries are useless to me -- I don't trust them and won't run them -- but sources in *any* language (well, maybe excepting Forth) are potentially of interest to me, and I save them. -- Larry Campbell The Boston Software Works, Inc. Internet: campbell@maynard.BSW.COM 120 Fulton Street, Boston MA 02109 uucp: {alliant,think,wjh12}!maynard!campbell +1 617 367 6846