Module naming and declarations
Tab Atkins Jr.
jackalmage at gmail.com
Wed May 1 08:28:57 PDT 2013
On Wed, May 1, 2013 at 7:18 AM, Kevin Smith <zenparsing at gmail.com> wrote:
> - Dave, your argument that URI's as a naming mechanism is a "failure"
> cherry-picks cases where URIs were obviously overkill. You have not shown
> that URIs are overkill in this situation. In order to do so, you would need
> to posit a centralized naming authority, or a dependency management system
> (e.g. package manager) with attendant dependency metadata. Or else you
> would need to show that for any reasonable selected subgraph of the global
> module (really package) graph, the chance of name collision is near-zero.
Or, we can simply point to real-world examples of large-scale custom
namespaces, where there is no central authority, the chance of
collision is very low and can be largely predicted and worked around
ahead of time, and in the case of collisions, the author can manually
work around the issue without too much difficulty.
See: jQuery modules, python modules, and many others. Some of these
*have* a central distribution authority which prevents name conflicts
for people using it, but it's optional to use.
Central naming authorities are only necessary if you need complete
machine-verifiable consistency without collisions. As long as humans
are in the loop, they tend to do a pretty good job of avoiding
collisions, and managing them when they do happen.
More information about the es-discuss