# call for comments on base URI issue

**URL:** <https://discourse.citationstyles.org/t/call-for-comments-on-base-uri-issue/969>\
**Category:** CSL Development\
**Created:** [March 7, 2011, 2:28pm UTC](https://discourse.citationstyles.org/t/call-for-comments-on-base-uri-issue/969 "2011-03-07T14:28:48Z")\
**Posts on this page:** 7\
**Page:** 2

<div class="post-metadata">

**Author:** ![Ron\_Jerome](https://avatars.discourse-cdn.com/v4/letter/r/e480ec/32.png) [@Ron\_Jerome](https://discourse.citationstyles.org/u/Ron_Jerome)\
**Post date:** [March 18, 2011, 5:46pm UTC](https://discourse.citationstyles.org/t/call-for-comments-on-base-uri-issue/969/21 "2011-03-18T17:46:34Z")

</div>

Sorry I came to this party a bit late, but I’ve been on vacation…

The other thing to keep in mind about serving styles directly from github is  
their access rate limits. I ran into the 60 request/minute API limit, and  
I’d be willing to bet if millions of clients suddenly start to check and  
update styles, neither the github folks nor the clients will be too happy.

Ron.

---

<div class="post-metadata">

**Author:** ![Bruce\_D\_Arcus1](https://avatars.discourse-cdn.com/v4/letter/b/4bbf92/32.png) [@Bruce\_D\_Arcus1](https://discourse.citationstyles.org/u/Bruce_D_Arcus1)\
**Post date:** [March 18, 2011, 8:12pm UTC](https://discourse.citationstyles.org/t/call-for-comments-on-base-uri-issue/969/22 "2011-03-18T20:12:06Z")

</div>

Is that true for static pages as well, or just for hitting the API?

In any case, perhaps we can proceed in such a way that we can easily  
switch over to dedicated hosting if the need arises?

Bruce

---

<div class="post-metadata">

**Author:** ![Avram\_Lyon](https://avatars.discourse-cdn.com/v4/letter/a/a88e4f/32.png) [@Avram\_Lyon](https://discourse.citationstyles.org/u/Avram_Lyon)\
**Post date:** [March 18, 2011, 8:19pm UTC](https://discourse.citationstyles.org/t/call-for-comments-on-base-uri-issue/969/23 "2011-03-18T20:19:47Z")

</div>

This is why I thought that somehow the base URI really ought to be a  
subdomain of [citationstyles.org](http://citationstyles.org) – it’s the only way to move these  
hits off of [citationstyles.org](http://citationstyles.org) the website without committing to a  
single external hosting service.

[repository.citationstyles.org/chicago.csl](http://repository.citationstyles.org/chicago.csl) sounds good to me.

It would still be possible to use redirects to continue hosting at  
Github, or the subdomain could refer to a server [probably a vhost, of  
course] run by some other generous host (Zotero? Mendeley? Mekentosj?  
Colwiz? Pandoc?), so that the zero-revenue CSL organization doesn’t  
have to bear any costs beyond domain registration.

Avram2011/3/18 Bruce D’Arcus \<@Bruce\_D\_Arcus1\>:

---

<div class="post-metadata">

**Author:** ![Bruce\_D\_Arcus1](https://avatars.discourse-cdn.com/v4/letter/b/4bbf92/32.png) [@Bruce\_D\_Arcus1](https://discourse.citationstyles.org/u/Bruce_D_Arcus1)\
**Post date:** [March 18, 2011, 9:34pm UTC](https://discourse.citationstyles.org/t/call-for-comments-on-base-uri-issue/969/24 "2011-03-18T21:34:38Z")

</div>

> > In any case, perhaps we can proceed in such a way that we can easily  
> > switch over to dedicated hosting if the need arises?
> 
> This is why I thought that somehow the base URI really ought to be a  
> subdomain of [citationstyles.org](http://citationstyles.org) – it’s the only way to move these  
> hits off of [citationstyles.org](http://citationstyles.org) the website without committing to a  
> single external hosting service.

I’m not committed to keeping the existing [citationstyles.org](http://citationstyles.org) wordpress  
site though. I much prefer to do blogging through something like  
jekyll.

In any case, that’s a separate issue from the bigger questions here  
about the actual styles (and hopefully later, the editor).

> [repository.citationstyles.org/chicago.csl](http://repository.citationstyles.org/chicago.csl) sounds good to me.
> 
> It would still be possible to use redirects to continue hosting at  
> Github,

Does github really allow that?

> or the subdomain could refer to a server [probably a vhost, of  
> course] run by some other generous host (Zotero? Mendeley? Mekentosj?  
> Colwiz? Pandoc?), so that the zero-revenue CSL organization doesn’t  
> have to bear any costs beyond domain registration.

Right. I think if we do hosting off github, it makes sense to have the  
CDS deal with it, given it’s a non-profit, but has excellent technical  
infrastructure. That’s how [citationstyles.org](http://citationstyles.org) is currently hosted.

Bruce

---

<div class="post-metadata">

**Author:** ![Avram\_Lyon](https://avatars.discourse-cdn.com/v4/letter/a/a88e4f/32.png) [@Avram\_Lyon](https://discourse.citationstyles.org/u/Avram_Lyon)\
**Post date:** [March 18, 2011, 9:42pm UTC](https://discourse.citationstyles.org/t/call-for-comments-on-base-uri-issue/969/25 "2011-03-18T21:42:54Z")

</div>

If the millions of hits are not a problem, [citationstyles.org](http://citationstyles.org) could  
serve 302 redirects for styles that need to be picked up from Github,  
and 304 for ones that haven’t been changed since the last update. In  
general, style-consuming clients should be conservative in their  
requests and should, I hope, be able to send If-Modified-Since  
headers.

This would still be a meaningful load on the servers, but the actual  
file serving could be off-loaded to Github, reducing bandwidth usage  
and perhaps simplifying the infrastructure; [citationstyles.org](http://citationstyles.org) would  
still need some custom kit to serve this way, and it would have to be  
kept abreast of Github commits, so as to send 302/304 as appropriate.

But CDS, possibly with some donations to CDS for this purpose from the  
other CSL users, sounds like a perfectly reasonable home.

Avram2011/3/19 Bruce D’Arcus \<@Bruce\_D\_Arcus1\>:

---

<div class="post-metadata">

**Author:** ![Bruce\_D\_Arcus1](https://avatars.discourse-cdn.com/v4/letter/b/4bbf92/32.png) [@Bruce\_D\_Arcus1](https://discourse.citationstyles.org/u/Bruce_D_Arcus1)\
**Post date:** [March 19, 2011, 5:05pm UTC](https://discourse.citationstyles.org/t/call-for-comments-on-base-uri-issue/969/26 "2011-03-19T17:05:28Z")

</div>

Now there’s an idea 🙂

In any case, we still have some details to iron out, and they’re  
important details given the fact of dependent styles.

Bruce

---

<div class="post-metadata">

**Author:** ![Bruce\_D\_Arcus1](https://avatars.discourse-cdn.com/v4/letter/b/4bbf92/32.png) [@Bruce\_D\_Arcus1](https://discourse.citationstyles.org/u/Bruce_D_Arcus1)\
**Post date:** [March 19, 2011, 5:16pm UTC](https://discourse.citationstyles.org/t/call-for-comments-on-base-uri-issue/969/27 "2011-03-19T17:16:22Z")

</div>

One random thought of a possibility …

> In any case, we still have some details to iron out, and they’re  
> important details given the fact of dependent styles.

What if we simply split the process of indexing changes and doing  
notifications from actually serving the style files?

E.g. imagine a little json index file, created from a local git repo,  
that looked something like:

{ // hash is keyed by style id  
“[http://citationstyles.org/styles/apa.csl](http://citationstyles.org/styles/apa.csl)” : {  
“blob”: “apa.csl”, // the local file name in the git repo  
“title”: “Association of American Psychologists”,  
“short-title”: “APA”,  
“updated”: “[date-time]”,  
}  
“[http://citationstyles.org/styles/foo.csl](http://citationstyles.org/styles/foo.csl)” : {  
“depends\_on”: “[http://citationstyles.org/styles/bar.csl](http://citationstyles.org/styles/bar.csl)”,  
“title”: “Foo Style”,  
“short-title”: “Foo”,  
},  
“[http://citationstyles.org/styles/bar.csl](http://citationstyles.org/styles/bar.csl)” : {  
“blob”: “bar.csl”,  
“title”: “Bar Style”,  
“short-title”: “Bar”,  
“updated”: “[date-time]”,  
}  
}

So clients with large user bases just work with that, and then it  
makes no difference where the index files are hosted, or even where  
the repo is? And you’d never need to hit the github API.

Bruce

[Previous page](https://discourse.citationstyles.org/t/call-for-comments-on-base-uri-issue/969.md?page=1)