[ccan] Moving modules to vanity naming?

[david at lang.hm]

On Wed, 3 Nov 2010, Rusty Russell wrote:

> On Wed, 3 Nov 2010 12:32:57 pm Alexander 'hatred' Drozdoff wrote:
>> In this case, I think, more good put 'author' first - with this simple 'ls' group author's
>> packages :-)
>
> I think that group-by-name is what David wanted to avoid, but I don't really
> mind.

while it wouldn't be my first choice, a single layer with author-module 
would be better than a two layer author/module, at least with the first 
you can see the module names in an 'ls', and if you care enough you can 
send the ls through a sort to sort by module name.

but what I had been thinking was that instead of author/module do 
category/module-author (or author-module)

by the way, one thing to remember with git is that in one repository (with 
one copy of the data) you can have multiple heads, so you can have a main 
ccan head, but also have a ccan-bsd head that only contains links to bsd 
compatible licensed stuff.

this is meaningless if you really expect people to clone individual 
modules, but I think that until this gets quite large people are more 
likely to clode everything and then look at the parts they need (the 
overhead of doing multiple clone commands will be higher than the overhead 
of pulling the extra pieces)

David Lang

>> Also, is it possible to add 'tags' functionality to CCAN? Add some new field to manifest file
>> (_info), like:
>>   Tags: memory, image processing, crypto
>>
>> And use this for navigation on site. This can solve trouble with navigation via modules with same
>> (or in one subject areas) functionality. Use this only for navigation on site.
>
> Certainly, and there's nothing stopping people starting that now.  But maybe
> we're better off just writing clear descriptions and letting Google sort it
> out?  We already extract that for the module web page...
>
> Cheers,
> Rusty.
> _______________________________________________
> ccan mailing list
> ccan at lists.ozlabs.org
> https://lists.ozlabs.org/listinfo/ccan
>