[Tutor] Best Python Editor

Alan Gauld alan.gauld at btinternet.com
Mon Jun 15 18:29:42 CEST 2009


"Emile van Sebille" <emile at fenx.com> wrote

> Anyone know of any studies comparing text based vs GUI IDE based code 
> development?  As I recall, programming productivity is measured in 
> LOC/day and last time I noticed it seemed to be a very small number.

When I started in "software engineering" (c1985) the typical value
was  between 8-10 lines of working code per person per day.
After PCs (and Unix workstations) became the norm the figure
slowly rose to around 12-15. With the latest tools I suspect it
might be getting close to breaking 20. (On small projects it will be
much higher, this is an industry average remember) But i haven't
seen a published figure for at least 5 years, if not more!

> I'm wondering if there might be documented benefits to migrating from my 
> horse and buggy.  :)

Most improvements in productivity were down to things like seeing
more than 24 lines of code at a time and faster navigation, as well
as much faster compilatrion speeds (who remembers the days of
limits on the number of compiles you could do per
day - usually 3 or 4 - or the days when a full build was measured
in hours, sometimes days (I worked on one project where we could
only do a full build on a Friday night because it didn't finish until late
Sunday afternoon...).

The IDE doesn't add a huge amount of improvement, and the real issues
holding down the figures are those that Fred Brooks called the
"essential problems" - like understanding the requirements, defining
an architecture etc - as well as regression testing etc. Even with test
automation it still takes us 6 weeks to run a full regression test of
some of our bigger projects. (Down from 16 though! :-)

As for tests comparing text based editing v mouse driven those
are very inconclusive and, I suspect, reflect the bias of the tester.
Back when the Mac first came out(1984?) Apple showed that
WYSIWYG word processors were much faster than older text
styles(Wordstar etc) but then Wordstar showed that text based
documents had fewer errors and gained students higher marks...

I've certainly seen comparisons going both ways for programmer's
editors. The one constant is that for pure text editing vi is usually
fastest in the hands of an expert(*) even compared to emacs. But
programmers do so much more than simply editing text!

(*)The old DOS editor Brief ran it a very close second. Brief morphed
into an Opensource GUI editor after Windows 3 came out (and Borland
bought Brief)  but I've forgotten the name and haven't heard of it for 
years.
....Wikipedia to the rescue - CrisP was the name and it was Unix only
apparently. Then did the very unusual thing of going from Opensource
to commercial! Its still going with a release in 2008.

Ahh, nostalgia :-)

Alan G. 




More information about the Tutor mailing list