15:58 < cj> lluis: since my friend isn't online to bounce ideas off of, how
'bout I ask you? :)
15:59 < lluis> cj: heh, ok
15:59 < cj> a first pass would be a web form that takes some information about
a package and stores it to a database. those data are exposed via
as an xml file, and all entries are listed on a separate page as a
16:00 < cj> http://someregistry.org/register_product.cgi, /list_products.cgi,
16:01 < cj> the data stored in relation to the product would be version,
package type, build type, license, distribution uri, author
16:02 < cj> the IDE would add another tab next to 'browse', 'pkg-config' and
'gac' that would query /list_products.cgi and present them in a
16:03 < cj> I don't want to over-engineer, since I don't have a lot of spare
time to implement. first pass = dumb but working
16:03 < lluis> well, if we are going to have server side logic, we better use a
16:04 < cj> sounds reasonable, but I don't have time to ramp up on that right
now. how 'bout I make the cgi's interface reliable so a web
service could be built to replace it?
16:05 < mhutch> cj: clean urls please :)
16:05 < cj> mhutch: define 'clean' :)
16:06 < mhutch> no ".cgi" for a start
16:06 < cj> okay
16:06 < lluis> cj: what you plan to use to implement that?
16:07 < cj> lluis: perl. I'm good at it and can fix things without having to
ask stupid questions like "how do I construct an HTTP request
16:07 * mhutch cries
16:08 < cj> enough from the peanut gallery
16:09 * mhutch starts throwing peanuts at cj
16:11 < lluis> cj: well, the server side implementation is not so important
16:11 < lluis> as well as we have a client c# library that encapsulates it
16:11 < lluis> I mean as long as
16:12 < mhutch> IMO it would make sense to do it in a language that users (i.e.
potential contributors) are likely to know
16:12 < mhutch> but maybe that's just me
16:12 < cj> mhutch: I promise that I'll write it in c# after I get it working
16:13 < cj> mhutch: it's hard for me to get anything done, though, if I have to
learn how to do it while I'm doing it. I enjoy re-writing my perl
code in c#, though. that's how I started learning it in the first
16:13 < mhutch> cj: I'd be happy to provide pointers
16:14 < lluis> mhutch: I don't think it is so important for the server side.
16:14 < mhutch> cj: some things would be easier to do directly in C#, for
example, web services in C# are near-trivial
16:17 < lluis> cj: hmm, in fact in my implementation proposal the reusable
components are just add-ins, so we can just use the existing
add-in repository format
16:17 < cj> lluis: is there one? I'll google it.
16:18 < lluis> cj: I don't think it is documented. Add-in repos are just as set
of files which can be generated using the mautil tool.
16:53 < lluis> cj: if you are interested, I can send you the library repository
client I wrote when I worked on that on the hack week
17:11 < cj> lluis: sure, thanks. I'll use it as reference as I poke.
Day changed to 09 Jan 2009
03:10 <@cj> okay... now for coding...
03:20 <@cj> this is the part where I use my production web server to develop
code. everybody watch!
03:29 <@cj> http://repository.colliertech.org/
03:29 <@cj> tada
03:29 <@cj> I crack myself up.
04:13 <@cj> http://repository.colliertech.org/register/
04:13 < SeveredCross> Neato.
04:14 <@cj> yeah. doesn't store your data yet, but *shrug* it's >0
04:15 < SeveredCross> What's the idea, to have a repository of info for
Mono-related softwares that aren't packaged in distro
04:15 <@cj> no, to allow people to reference the packages registered here
through various tools, such as visual studio or monodevelop
04:16 <@cj> http://wp.colliertech.org/cj/?p=301
04:16 <@cj> http://wp.colliertech.org/cj/?p=307
04:17 <@cj> at amazon, we had a 'cradle-to-grave' registration tool that allowd
folks to create source control repositories, fire off builds,
request and perform code review, etc.
04:17 <@cj> I don't want to get that fancy, but it inspired me to make
something that helps to oranize all of these pieces of code we have
04:18 <@cj> organize
04:20 <@cj> pkg-config is useful for code you've got on your system, but the
process of grabbing, installing and then finally referencing is a
lot of overhead. if you could do gmcs -r:repos://foo bar.cs
-o:baz.exe, that would really reduce the headache
04:20 < SeveredCross> Cool beans.
04:20 < SeveredCross> Sounds like a worthy project. :D