On Tue, 19 Jun 2018 21:34:05 -0600, Joe Pfeiffer
Post by Joe PfeifferPost by James KuyperPost by Joe PfeifferPost by Lynn McGuirePost by Lynn McGuire"The Soon-to-Be-Extinct Embedded Software Engineer"
https://www.designnews.com/design-hardware-software/soon-be-extinct-
embedded-software-engineer/39152617858743
Post by Lynn McGuire"Embedded software engineers of the future will have a very different
skillset from their traditional predecessors. Theyll know how to call
an API to make the hardware do something, but they wont know why or how
it does it."
Um, no.
Lynn
I remember hearing, in the late 80s, that software development,
as a job, would soon disappear on account of the projection that
applications would be, more or less, automatically generated by the
interested parties. This a bit like religious fundamentalists and their
ilk, predicting the Rapture "real soon now" on a regular basis.
So far as I can determine, this prediction has been made on a regular
basis ever since FORTRAN was first released....
I've heard that something similar was said about FORTRAN itself, by
comparison with assembler.
Maybe I wasn't clear -- that's what I meant. I can't cite the original
FORTRAN paper (in CACM?) since I can't find the copy I used to have, but
my recollection is the claim was made in that paper. It also (IIRC)
made the claim that it would virtually eliminate programming errors, and
the professor (this was when I was in grad school, roughly 1980) made
the comment that "virtually" means "doesn't". That's a definition that
has stuck with me over the decades.
Of course there's no question that HLLs have eliminated a great many
low level programming errors in practice. Which unfortunately appears
mainly to have freed up a lot of programmer's time so that they can
concentrate on making high level programming errors...
There's an old saw that programmers generally write about the same
number of LOCs per unit time, no matter what language they're
programming in (the number varies by programmer, of course). There
appears to be at least some truth to that.
It's been my observation that the number of errors produced per unit
time is fairly constant as well.