Calculators, Binary Hybrids, and UNIX History

I wanted to add a calculator to my leatherman but I never ever want to write a parser. The following is what ensued.

I’ve ended up in situations where I have my leatherman handy but bc(1) is not installed. I very briefly considered implementing my own calculator, but every time I immediately consider the actual parsing of expressions and discard the idea. I hate writing parsers!

Recently I considered that maybe there’s some weird linker trick I could do such that the binary would have a statically liked bc(1) appended. I asked some coworkers and basically the answer was that I could compile my leatherman as a shared library that gets linked into a hacked up bc(1) that would know how to call out to the leatherman entrypoint. This is a bridge too far for just using a calculator.

Side note: I expect at some point to make some rust based tools and having them built into leatherman in this fashion would be pretty cool, so I expect that to happen in the next year or two.

I checked to see if busybox has a bc(1) implementation and verified that it indeed does not. But it does have dc(1), the reverse polish notation calculator that preceeded bc(1):

$ echo "11 3 / p" | busybox dc
3.66667

(That’s push 11 and 3 onto the stack, divide the first two numbers on the stack, and pop/print the number on the stack.)

My coworker Jeremy Leader pointed out that bc(1) actually used to be a frontend for dc(1). I happened to notice that when reading the manpage for GNU bc(1) today:

   DIFFERENCES
       This version of  bc  was  implemented  from  the
       POSIX  P1003.2/D11  draft  and  contains several
       differences and extensions relative to the draft
       and  traditional  implementations.   It  is  not
       implemented in the traditional way using  dc(1).
       This  version  is  a single process which parses
       and runs a byte code translation of the program.

My hunch was that “the traditional way” was some yacc based parser. I did some digging and verified that, as far as I can tell, the original bc(1) implementation, indeed yacc, was authored by Ken Thompson and the late Dennis Ritchie around midnight on Wednesday the 14th of May, in 1975. (Note that the commit metadata shows a different timestamp, but it’s in Pacific Time and Bell Labs is on the east coast.)

---

I found this a fun little excursion. At the minimum, I know dc(1) is at the ready even in the rare situation that bc(1) is not (it happens!) On top of that it’s cool to be able to dig so deep into the history of your industry that you can find this kind of thing.

---

I think a natural book to recommend along with this is UNIX: A History and a Memoir. I haven’t read it yet but all the reactions I’ve seen about it are positive.

Another good, sorta related book is Coders at Work. I read it many years ago but really enjoyed it.

Posted Mon, Feb 10, 2020

---

If you're interested in being notified when new posts are published, you can subscribe here; you'll get an email once a week at the most.