Ciaran McCreesh <> posted
20090506223757.106ca...@snowcone, excerpted below, on  Wed, 06 May 2009
22:37:57 +0100:

> If a question asks "what's wrong with this code?", you know there's
> something wrong and you can spend time researching to find out what it
> is. But people need to be able to recognise mistakes even when they're
> not looking for them, and to know when something's wrong even if they
> haven't been told to find the mistake -- being able to do this requires
> having a good immediate knowledge of certain parts of the material.

You're right, but that's where the history comes in.  If the most recent 
code/ebuilds has/have been crap, needing lots of corrections, etc, they 
probably need some more pre-dev mentoring.  If it's good quality, well, 
perhaps they're ready to go into apprenticeship, aka actively mentored 
new dev.  After all, the commits from current devs aren't always perfect, 
as can be seen by the comments on the commit-feed from time to time.

Plus, as I said, with a pre-arrangement, it's possible to do email 
reasonably close to real-time as well, close enough they'd not have time 
to look it up unless they had /some/ idea what was going on.

But it's also possible to structure those questions a bit differently.  
Provide several samples of code, some of which have problems, some of 
which don't.  Ask them what they'd change if anything, and why.  Throw 
some in from the commit feed which might have minor stuff, plus a couple 
of known critically wrong examples.  Tell them you'll expect a rough 
"what's wrong" (for the group of several samples) in say 10/20/whatever 
minutes, and fixes within an hour/2/whatever.  There's no reason that 
can't be done via email, and throwing in some live commit feed action 
might make it a bit interesting. =:^)

Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman

Reply via email to