On 2013-12-02 11:18:25 +0100, FRIGN wrote:
> You can present the benefits of compile-time-configuration even more
> and make even clearer, that compiling nowadays doesn't take much time
> anymore (at least for me, not having to learn yet another config-
> interface for package xy).
From my experiences talking to people about the matter, there are three
main hurdles (that don't necessarily need to be overcome, because I
think they're mostly just things that people have to get over):
1. Viewing compilation as "dirty" on a system that is generally
maintained by a package manager. The solution is, of course, either to
stop caring (because you're putting it in the right place according to
the FHS, so it doesn't matter), or to roll your own package. In some
situations the latter is not ideal if you're pulling down from git.
2. Being intimidated by the C spec, not that you really need to know
much about it just to modify config.h, but just the concept is
intimidating for them. The solution to this is for them to actually try
doing it, probably.
3. Viewing the ability to execute arbitrary code as part of the
configuration as a bad thing. I think people have got into this
mentality that configuration files must be sanitised-by-design, that
they should only be a series of on and off switches and data structures
that could not possibly affect the actual execution of the main program.
Not that we typically do anything like this from config.h... but I think
some people still don't like it. The solution is to stop caring, this is
a non-issue, this is the sort of mentality that got DRM created.
- application/pgp-signature attachment: stored
Received on Mon Dec 02 2013 - 17:27:41 CET