[prev in list] [next in list] [prev in thread] [next in thread] 

List:       mifos-developer
Subject:    Re: [Mifos-developer] Linux v Windows
From:       "Jim Kingdon" <jkingdon () grameenfoundation ! org>
Date:       2006-12-13 17:04:08
Message-ID: 9DD845C1ED0D5D40B4B56DF5A4B1EB0E01190A77 () gfmail ! gfusa ! org
[Download RAW message or body]

> Here's a reference for the kinda simple way to avoid this bug
> http://java.sun.com/docs/books/tutorial/essential/io/file.html

Not sure I agree that the various techniques in that page are
simple, or desirable.
Instead, I'd say the desired solution, at least most of the time,
is to use / (not \) on all platforms.  Windows has supported this
back to the MS-DOS 2.x days (the only part of Windows which doesn't
is COMMAND.COM, which isn't relevant here).

> Furthermore, I think it's clear, given the history on this project,
that 
> we need to do some pre-compilation checking for this
sub-sub-sub-species 
> of bug, and make the build break when this occurs. And it may even be 
> nice to throw in a few test or two.

Sounds like a good idea.  Given the various places that paths appear,
it might entail some work.  But I'm all for someone taking a look
at this.

> However, let's not forget that one of the important goals is bug-free 
> software. These errors are the result of poor coding practices and
need 
> to be corrected, preferably by the person who introduced the problem
in 
> the first place.

Yeah, if we can write tests, then this will happen naturally.
If we don't catch it at the time, our chances of tracking down
the original author and motivating them to fix it go down, so
this is yet another argument in favor of trying to catch these
things with tests.


[prev in list] [next in list] [prev in thread] [next in thread]