Unix: A History and a Memoir, by Brian Kernighan (opens in new tab)

Since history belongs to winners, if it wasn't for the accessibility of old conference papers and computer manuals, that would be indeed the only one we had to believe on.

[0] https://en.wikipedia.org/wiki/Wollongong [1] https://en.wikipedia.org/wiki/Royal_National_Park

I see it from the perspective of someone working at a fast-moving, Silicon Valley-based software vendor. In that context, I don't see a lot of appreciation for history, in the day-to-day basis.

But it's not because people intrinsically don't care about it. It's more about our daily lives. We not only have to do what we all have to do, but we also have to keep up with the latest if we want to survive here. That line of thinking constantly pushes us forward, and doesn't reward looking at the past.

If history exists, but nobody reads it, then are we not still lost?

Glad it has finally arrived!

Blaming (or crediting, however you want to think about it) GNU for proliferation of Unix is anachronistic. RMS has repeatedly said he doesn't care for the design of Unix, but he chose it for the ease of implementation. GNU wasn't even bootable as a stand-alone OS before 1990s and certainly not production ready until Linux was. Using GNU utilities in proprietary Unices was popular, at least since the 90s, but I never heard anyone consider them a "killer app" for Unix.

Computing is absolutely full of widely deployed working technologies that enable people to get work done all day but that have rough edges. For every one of these, there is somebody saying that you should use <pet technology> instead. Usually with missionary levels of zeal. And yet at no point do they seriously address why people might have good reasons for adopting the allegedly inferior solution.

RMS worked on EmacsLisp.

Unix was in development in 1969, had a manual released inside AT&T in 1971 and was announced publicly in 1973. The first source license was sold in 1975.

VMS was released in 1977 as VAX/VMS on the VAX series of computers. Before VMS the DEC hardware ran various other operating systems such as RSX-11, TOPS-10, and optionally AT&T Unix (which was developed on the PDP series, first the PDP-7).

If we take ENIAC [0] as the first computer (1945), then our industry is 74 years old. Let’s compare this with another tech industry – the aviation industry. The Wright brothers flew in 1903 [1]. 1903 + 74 = 1977.

I was a kid in the 1970s, fascinated by aerospace and aviation. There was a massive amount of aviation history around (I still have some of the aviation history books I bought then). So I think there has been plenty of time for the history of computing to appear. Even if we added another 15 years to shift the start of the history of computing (software engineering) at the development of COBOL [2] (approx the same as FORTRAN or LISP), there has still been a lot of time.

So I don’t think it’s industry youth / age alone.

[0] https://en.wikipedia.org/wiki/ENIAC

[1] https://en.wikipedia.org/wiki/Wright_brothers

[2] https://en.wikipedia.org/wiki/COBOL

Brian Kernighan mentions in the book that awk provides "the most bang for the programming buck of any language--one can learn much of it in 5 or 10 minutes, and typical programs are only a few lines long" [p. 116, UNIX: A History and Memoir]. Also keep in mind Larry Wall's (inventor of Perl) famous quote/signature line: "I still say awk '{print $1}' a lot."

More background on awk from Brian Kernighan in a 2015 talk on language design: https://youtu.be/Sg4U4r_AgJU?t=19m45s

In a way, awk, like sed, cut(1), paste(1), comm(1) is closer to the original Unix design philosophy of pipeline composition of simple, single function tools. cut, paste, and comm are simple, single function. sed is programmable, and awk even more so, but they're still optimized for being used in pipelines. Perl still has lots of options designed for pipeline usage (-p, -n, -a, -0, -F), but was always capable of doing general purpose programming, and python is primarily general purpose programming.

Personally, I've never learned sed, cut, paste, or comm; the only traditional text processing tools I use are sort & uniq. I know awk, but the one time I wanted to use it in the last 5 years in an official script, I was shouted down by confused youngsters. I use perl for one liners, and python for any nontrivial scripting.

I once wrote code with curl, awk, sed, jq, xargs and in less than 30 lines implemented something that would have taken about 200 lines of go code. I did it too in less than an hour when I was told it would take a day. It was urgent and I needed the result fast.

There are different kind of hammers, different kind of nails and screws. Know your tools, use them when it's called for. You can use a wrench as a hammer, doesn't make it a great one.

For a long while I maintained a program that took the compiler output (intermediate form) from one toolchain, and converted it to another (intermediate form).

Why awk? It was faster, and more readable, than any of the alternatives, whilst being much much shorter.

Youtube-dl and awk to grep out of a YouTube channel.

If I were doing it again I would use youtube-dl -o to just print the base64, instead of "manually" renaming it to that with ls>awk:mv>sh.

I only see these items plus another cooler and a Macbook stand.

I can’t remember a single mainstream title that hasn’t had an ebook version for preorder and first day sales. This includes “Bad Blood”, “Super Pumped”, and “Catch and Kill”.

I have stopped buying paper books over past several years as they simply accumulate moisture and mould in the climate I live in and make rooms smell musty. It is incomprehensible to me why primarily textual books would be published without and ebook version these days.