[Exherbo-dev] Tags vs. categories
bryan.ostergaard at gmail.com
Fri Jul 11 10:27:28 BST 2008
On Fri, Jul 11, 2008 at 12:55 AM, Ciaran McCreesh
<ciaran.mccreesh at googlemail.com> wrote:
> On Thu, 10 Jul 2008 11:26:47 +0200
> "Bryan Østergaard" <bryan.ostergaard at gmail.com> wrote:
>> filesystem layout. The way I see tags are much more like the
>> DESCRIPTION variable in exheres.
> Right. If tags are just a metadata key containing a set of words,
> adding in tags is trivial. You'd do something like this in the exheres:
> TAGS="cplusplus python pdf"
> and be able to search with it using inquisitio as normal.
> Or we can be more structred:
> From a package manager perspective, they're largely equivalent.
Structured is probably better as it could allow more specific searches.
> But the question is, are tags merely a search / browse utility? Or are
> they used in identification too? Both of the above are only the former.
Indeed. Searching is fairly obvious while identification has some
problems that I don't see any obvious solutions for.
>> Package aliases are however as interesting to me as they've always
>> been - allowing the apache httpd to be known as both net-www/apache
>> and net-www/httpd and making sure the package manager knows it's the
>> same package would be fairly cool imo. But that only makes it easier
>> to find the package if you know (one of) its name. Tags (as applied to
>> searching) would allow you to search for packages based on what
>> features the packages provides.
> The only problem with package aliases is persistence. If you install
> apache, and then someone aliases httpd to apache, and then something
> deps upon httpd, how do we know that the installed apache is also known
> as httpd?
Just wing it the way Gentoo does package moves and implement some icky
NDBAM update as part of --sync :P
Slightly more seriously - I'm aware that we need to solve the
persistence problem of course and I don't want to push for aliases
without a proper solution for that.
More information about the Exherbo-dev