[erlang-questions] Style wars: junk comments

Richard O'Keefe <>
Thu Sep 13 02:11:49 CEST 2012


On 13/09/2012, at 2:13 AM, Alexandre Aguiar wrote:

>  escreveu:
>> When you are coping with over 100,000 lines of code,
>> tools that help you *NOT* read are desperately valuable.
> 
> Note that your case report is perfect to demonstrate that printing should *not* be routine.

Bad logic.  Yes, printing the *whole* thing should not be routine.
It does *NOT* follow that you should not print a small number of
modules.

> Printing selected limited portions is what helped you.

Exactly so, but we are talking about
 - the entirety of a 'module'
 - up to 100 pages.

> At a 100 lines per page, handling a 1,000 pages (probably unindexed) is a hard task.

50 to 60 lines per page would be more realistic.

Why on earth do you say "probably unindexed"?
Part of the point of printing is to use listing programs
that generate aids to navigation, like headers that
identify module and function, tables of contents, and
indices.

The 2009-08-23 draft of the COBOL standard is 910 pages.
I have in fact been reading it on an iPad, but the whole
time I've been wishing I had a paper copy.
The SQL standard dwarfs it.

> :-) And, of course, a waste.

Why "of course"?  Printing something is a waste if and only if
the cost in time, paper, toner, environmental effects, &c
exceeds the benefits.  If it gets read, especially if it gets
read more than once, it is not a waste.  If you can read it
more comfortably than you can on screen, it is not a waste.
The iPad 3 has a lovely screen, but even that is still not as
good as paper.  If printing saves eyestrain (and for me it
often does), it's not a waste.

In the case of the 100,000 lines of code I mentioned, every
line has been printed at least once, and it has _always_ paid
off *for me*.

The relevant factor here is probably how much you can read
before it gets out of date




More information about the erlang-questions mailing list