Source Code Distribution From Printouts to Github

2020-10-10 / geek / unix / history / 100daystooffload

Source code distribution has changed over the years. Today we all love (hate?) git, github and friends, but, believe it or not there were ways to distribute source code even before the Internet. In fact, this was the world in which the GNU Public License was created. Below are a few of the ways I’ve gotten/transferred source code through the years, in something like chronological order

Post 31 of #100DaysToOffload https://100daystooffload.com/

Print Outs

Paper. Yes, paper. My first memories of source code (late 70s) include source listing of BASIC programs printed on ASR-33 Teletypes. I also have memories of a large stack of “green bar” printouts at home (again, c.a. 78 or 79) that I think where the printouts of Emacs (TECO emacs, the original, before GNU) that my brother probably brought home from Carnegie-Mellon where he went as an undergrad.

Paper Tape

The ASR-33s had paper tape punches on the side. You could punch your programs onto tape and read them back in. It was 8-bit ASCII. At one point, I think my brother could read the ASCII by looking at the tape.

Cards

Early in college (80, 81) I actually punched cards for a class or two. You could duplicate cards. No cheating on that programming assignment now !

BYTE Magazine

Gamers of yore bought BYTE Magazine (and others), in part to get the source code listings that they could then type in.

ACM, Knuth

And of course there were textbooks such as Kunths (to this day unfinished) “The Art of Computer Programming” and ACM publications. Steve Wilhite used an algorithm published in ACM as part of GIF…of course anything printed was in the public domain, right? … well, no. We were all still learning about and in some cases defining the legal status of software then, right along with GNU/FSF.

USENET News

In the late 80s, MIT project Athena periodically posted the source code to X-Windows to the USENET news group comp.sources.x. If you were on a USENET site that “took a feed” of that newsgroup (usually over modems not faster than 9600baud), to get the latest source to X, you watched the news group (using readnews, rn or some other “newsreader” program), saved off the individual pieces from each post, reassembled and then extracted the code. The (large) distro was compressed and broken up into many pieces, and re-assembling them often involved the use of uudecode, shar, patch, tar, and programs to decompress (pretty sure gzip was not a thing yet). Of course, you had to grab them before your news site expired the articles (spool space was limited, and space was shared with rec.humor.funny and other essential news groups). Heaven help you if you missed one piece. No bit-torrent for you to find the missing pieces. You got one shot from one source in a limited time window.

UUCP

UUCP being “unix to unix copy”, you could literally issue a copy command of source known to live on a remote system something like

            uucp cbosgd!/archive/hdb-uucp.tar.gz hdb-uucp.tar.gz

with multi-hop paths being possible…if you knew the topology (pathalias !) and the filenames.

mag tape

Then there was the time I was in Boston (visiting Encore computer?). I took a blank 9-track tape, arranged to stop by MIT, was given a personal tour of the labs, show the locked room where the all-important KDC (forerunner of Microsoft’s domain controllers) was kept, and got a copy of the latest Athena source (X, Kerberos, Hesiod, etc.) dumped to tape. This was getting software, the easy way.

DECUS tapes

Tapes were a thing for quite a while. early on users of DEC computers created and distributed “decus tapes”, which were user-contributed collections of software. If the VAX/VMS bulletin board system I created in FORTRAN-77 around 1983 exists anywhere in the world, it’s probably in the archive of DECUS tapes. This is one example of collections of user-generated software that grew into the software archives we came to know later. I think the IBM SHARE user group had similar collections. The programs (games, etc) distributed with BSD Unix formed another set. DistroWatch is one web site in this vein today.

Floppies

With the advent of the PC (early, mid 80s) floppies became a thing. I purchased a copy of the Mark Williams C compiler for DOS just to get the public-domain source to MicroEmacs which I then ported to the Amiga. Of course, with floppies came boot sector viruses and the security wars were on. Later issues of BYTE sometimes shipped with floppies of source so you didn’t have to type it in.

Email

Yes, you could always use tar, shar(1), uuencode(1) and friends to email source code. Email has been a workhorse for something like 40 years, not just for communication but as the file transfer method of last (first?) resort.

CompuServe (and other) Forums

I ran (“SYSOPed” the CompuServe Unix Forum for a few years. I think we/I may have invented the file suffix “.tgz” to stand for “gzip tar” archive (because the forums could not deal with multiple suffixes, as in .tar.gz). We had source code available for download. The Forums were just CompuServe’s version of the local dial-up BBS and many other similar message/download systems.

FTP

As Internet connectivity started becoming common (at least in universities, government and some larger corporations), a number of “well known” FTP sites where created (often at well connected UUCP sites) that collected and offered source downloads. UUNET and WUSTL (Washington Univerity at St. Louis) being some of the larger ones (there is still a version of FTP server software, wu-ftp available for Linux)

Source Code Control Systems

As Internet connectivity became a thing, source code control systems such as cvs, svn, mercurial and friends started being used on local networks and it became possible to use them over the Internet. They also solved a lot of the collaboration, version control and release management issues that are important for projects of any size.

SourceForge

Sourceforge gets honorable mention as a precursor to github and friends. In the early 2000s it hosted many open source projects. I initially hosted NCAT there which became the Center for Internet Security’s Router Audit Tool.

“My name is Legion, for we are many”

Git is a good thing. Really. Today we have github, gitlab, dropbox and thousands of other options for getting and sharing source code, some of which even work reasonably well.

The world has changed. It really is a better place. No more trolling the USENET News feed spool. No more trips to MIT, tape in hand. No more hoping ftp.uu.net has the latest version. Go forth and hack.