Issue with Jekyll

[karthik]

Works fine locally, remote renders stub in ropensci/blog/ but full link fails because pretty url cannot translate to + .html.

I moved the latest post to _drafts/ but if anyone wants to fix, move it to _posts and try again.

As Scott pointed out, we also need a sandbox for the website so we are not always discovering these in production.

[cboettig]

Um, seems to be working fine for me I think.... weird. See:

http://xsede.carlboettiger.info:4000/blog/2016/01/04/researcher-survey

I just cloned and ran:

docker run --rm --label=jekyll --volume=$(pwd):/srv/jekyll -it -p 4000:4000 jekyll/jekyll bundle exec jekyll serve

and it works without the .html. Maybe weirdness is on the server side with apache rewrite stuff?

p.s. um, doesn't look like you committed the move to drafts, so I think the latest copy is still in _posts.

I could set up a quick travis thing that would publish the site to, say, the gh-pages branch where we could preview it?

[karthik]

sounds good.

[karthik]

I guess I didn't push that change to _drafts

[karthik]

Can one of you (@sckott, @cboettig) help get this post on the website? Perhaps changing the date to today. Otherwise it loses its value. We also need a better system for this. If Carl's new solution to trigger a build and push to gh-pages is working, perhaps we have something other trigger that watches gh-pages and pushes that to our host?

On a related note, a small orange seems to be going south in their service. So after our plan ends this year, it might be worth moving all over to some other service. While plain S3 is nice, we do get some other sugar like email forwarding, easy config for subdomains etc without dealing with more webops. But open to ideas here since the rest of you know more about cloud services than me.

[sckott]

I'll take a look

[cboettig]

sounds good. Happy to help with the deploy but not really sure what I can do to be useful. I believe the recipe I put in the travis file works; but I don't think I have access the ASO service to actually do the deploy.

Sad to hear about ASO. If the site is fully static, I think we should just move it to gh-pages (fast CDN, great up time, one less moving part in the deploy process given that we're already in github/jekyll). Or Amazon S3+cloudfront if we want to keep https.

Then I think we can find separate solutions for email and domain forwarding. I have found Google DNS has been great, very easy to configure subdomains etc. They support email-forwarding as well but I haven't heard it.

[karthik]

After

bundle install
bundle exec jekyll serve --watch

I see this (i.e. it works):

Which is some bizarre fucking black magic because pushing the build to the site makes it fail. I vaguely feel like it has to do with some file that is ignored or not being rsynced. Wordpress doesn't seem that bad when dealing with stuff like this for every post (there isn't even code here).

[karthik]

Given there is nothing we can do to fix this, short of a major site overhaul (move everything to gh-pages, remove proprietary material etc) I'll just give up on this post. The twitter response will have to do for this round.