Re: [gentoo-dev] Ebuild Readability Issues

[Johannes Ballé]

Am Monday 02 December 2002 14:03 schrieb foser:
> Let's face it, using those longer names would make ebuilds bigger,
> uglier and probably faultier :the longer the varname, the higher the
> risk of mistakes. Given the amount of times these vars are used it would

That's not correct. The longer the varname, the greater the amount of 
redundancy. But redundancy in general does not CAUSE mistakes, redundancy 
allows for easier detection of mistakes. For instance, take a one-letter var 
name. A single typo can make it a totally different variable. When you take 5 
or 8 letters, a single typo would cause the variable to be undefined, and you 
could check for that. (I admit checking for unset variables may be difficult 
in shell scripts, but you get my point.)

> probably also increase the portage tree a few megs in size, i don't
> think that's acceptable.

That is an argument, although it shouldn't impact the traffic so much as rsync 
uses compression, anyway (unless it's turned off for some reason, correct me 
if I'm wrong). And regarding disk space, I wouldn't mind to have a few megs 
more reserved for the portage tree on my boxes.

> The best way to learn about ebuilds and variables inside them is still
> looking at other ebuilds.

Definitely not. You need a good reference, too. It's a major pain to have to 
search for one particular feature in the portage tree that you'd like to add 
to your ebuild. I tried this. In addition, if you use var names such as $P or 
$PN, looking at other ebuilds to find out what they mean can take you a long 
time.

Of course, if you write 2 or 3 ebuilds a day, long var names can become a 
pain, too. Therefore, I vote for short var names that are documented :-)

For that matter, "man 5 ebuild" looks fine. Maybe the dev howto should 
explicitly point to that manpage, since I, too, missed it.

Greetings.

-- 
Johannes Ballé <joba123@×××××.de>


--
gentoo-dev@g.o mailing list