[Neuroimaging] Website Update

[Ben Cipollini]

@Ariel I don't think that change solves anything different than simply
setting the baseurl to /nipy-jekyll/ (note: I would avoid the fully
qualified URL, it would unintentionally make a circle.ci version of the
site, or a locally built one without specifying an overriding baseurl, to
pull from an unexpected source.)

The problem is (at least as I see it), we need different baseurl values for
different build situations (gh-pages, circle.ci, local).


On Sat, Aug 22, 2015 at 12:12 PM, Ariel Rokem <arokem at gmail.com> wrote:

>
> On Sat, Aug 22, 2015 at 11:34 AM, vanessa sochat <vsochat at stanford.edu>
> wrote:
>
>> The "ugly" solution is to use different templates for the pages not at
>> the base url, and then you can just hard code the page one level up. For
>> example, in the header
>> <https://github.com/vsoch/nipy-jekyll/blob/master/_includes/head.html>:
>>
>> a path currently looks like:
>>
>> <script src="{{ "js/tipsy.js" | prepend: site.baseurl }}"></script>
>>
>> and for subpages, we would make a new header_subpage.html with a path
>> like:
>>
>> <script src="{{ "../js/tipsy.js" | prepend: site.baseurl }}"></script>
>>
>> But then when you change anything about the design, you have to change it
>> in two places. I started implementing this, felt sick to my stomach at how
>> awful the idea was, and completely converted everything back :) I think it
>> could be possible to use this weird language <http://liquidmarkup.org/>
>> so that all pages detect their level. For example, take a look here
>> <https://github.com/vsoch/nipy-jekyll/commit/2ac70c6dee2372be6676848aac9766459f4c57c2#diff-e921a390257abd15c1742300f2417b13R14>
>> at a solution I figured out using the syntax - but it fails in that the if
>> statement is hard coded.
>>
>> Also take a look here
>> <https://github.com/vsoch/nipy-jekyll/commit/c72756afad2bc4cb57d893f76e9dd4f425b79218#diff-4ea7a794e4d7b7503ac520689daeb20bL17>
>> - for the posts we can get the URLs programatically, but I wasn't sure how
>> to do that for "non-post" pages.
>>
>>
>>
> I am also thinking in this direction. Maybe we can have different baseurl
> configs for different things? Something along these lines:
>
>
> https://github.com/arokem/nipy-jekyll/commit/9f2b6be741b20b779321dc2739fc6c6efa1a3bac
>
> Still not working, but something like this might work. Once we settle on a
> design, I don't expect that we will have many changes there. I think that
> most PRs will be with text content, and not to things of this sort, so I am
> not too worried about the potential ugliness of having to change many
> things in many places for that.
>
>
>
>
>
>> On Sat, Aug 22, 2015 at 11:24 AM, Ben Cipollini <bcipolli at ucsd.edu>
>> wrote:
>>
>>> Oops, I spoke too soon. Sounds like circle.ci is very important. :)
>>>
>>> Vanessa, you had an idea for an "ugly" solution for the original issue?
>>> May I ask what that is? Looks like others are simply using --baseurl="" on
>>> the command-line.
>>>
>>> Thanks,
>>> Ben
>>>
>>>
>>> On Sat, Aug 22, 2015 at 11:19 AM, vanessa sochat <vsochat at stanford.edu>
>>> wrote:
>>>
>>>> Hi Ariel,
>>>>
>>>> Circle is needed to preview PRs.
>>>>
>>>> On your fork take a look at the individual blog pages - the path goes
>>>> to the base and it is not found. And then on circle since the baseurl is
>>>> "/" it would be quite broken looking.
>>>>
>>>> Right now adding a project would mean adding an entry to the data file
>>>> <https://github.com/vsoch/nipy-jekyll/blob/master/_data/packages.yml>,
>>>> and then copying a template page for it (for example
>>>> <https://github.com/vsoch/nipy-jekyll/blob/master/sdm.html>, you would
>>>> just need the header and to name the page corresponding to its tag in the
>>>> data file). The navigation, front page, and D3 are rendered from that file,
>>>> but I didn't see a nice way to generate an entire page for the package. The
>>>> reason I had originally chosen flask is because a person could just add a
>>>> line to a tsv file, and the application would be smart enough to make the
>>>> page if the user did not. I don't see a way to achieve that with these more
>>>> syntax based frameworks.
>>>>
>>>> For the content and more tweaks to the design - let's hold off on
>>>> jumping into that until we get this url issue worked out! I am totally game
>>>> for adjusting the design, layout, content for whatever people think is best.
>>>>
>>>> Best,
>>>>
>>>> Vanessa
>>>>
>>>> On Sat, Aug 22, 2015 at 11:04 AM, Ariel Rokem <arokem at gmail.com> wrote:
>>>>
>>>>> Hi Vanessa,
>>>>>
>>>>> Sorry for the delay in responding.
>>>>>
>>>>> On Sun, Aug 16, 2015 at 3:18 PM, vanessa sochat <vsochat at stanford.edu>
>>>>> wrote:
>>>>>
>>>>>> Hi Everyone,
>>>>>>
>>>>>> Here is an update to the site. I've ported it into jekyll
>>>>>> <https://github.com/vsoch/nipy-jekyll>, and tested the template on
>>>>>> iphone and computer (seems reasonable). I've also gotten it to render the
>>>>>> artifacts on circle.ci <https://circleci.com/gh/vsoch/nipy-jekyll/32>,
>>>>>> so we could preview PRs (Click "artifact" and then find home.html
>>>>>> <https://circle-artifacts.com/gh/vsoch/nipy-jekyll/32/artifacts/0/home/ubuntu/nipy-jekyll/_site/home.html>).
>>>>>> However, I've run into a bug with regard to absolute vs. relative paths -
>>>>>> when you click the demo link above, the "posts" not being at the base url
>>>>>> means they will not be linked or render correctly (for example, here
>>>>>> is a link
>>>>>> <https://circle-artifacts.com/gh/vsoch/nipy-jekyll/32/artifacts/0/home/ubuntu/nipy-jekyll/_site/code-syntax/index.html>
>>>>>> to a post on circle, and you won't see styling because the css/js links are
>>>>>> broken)
>>>>>>
>>>>>> The site works perfectly given that we aren't serving it at a
>>>>>> relative path, eg:
>>>>>>
>>>>>> jekyll serve --baseurl="/"
>>>>>>
>>>>>> And the challenge is finding a solution that will render locally, on
>>>>>> github pages, and circle.ci, and what we need, I think, is some kind
>>>>>> of liquid syntax that will detect when we are in a subfolder and render
>>>>>> appropriately. I've been messing around with it for 7 hours today and can't
>>>>>> figure it out, and I need to ask for help. Could someone else take a look?
>>>>>> Likely someone with more experience than me with Jekyll could figure this
>>>>>> out in a heartbeat (Ariel?) The plan right now, I'd like to propose, would
>>>>>> be to figure out this bug, get it onto a (test) github pages, tweak the
>>>>>> details of the content, make proper documentation, and then release to
>>>>>> official github pages.
>>>>>>
>>>>>>
>>>>> Why do we need CircleCI here? I am actually not sure what the issue
>>>>> is, but here's the site rendered through gh-pages on my fork of your repo:
>>>>>
>>>>> http://arokem.github.io/nipy-jekyll/
>>>>>
>>>>> I did also add a Gemfile to the repo, and that might matter: I believe
>>>>> Github pages serves jekyll through a call to "bundle exec jekyll serve"
>>>>> (see here: https://help.github.com/articles/using-jekyll-with-pages/
>>>>> ).
>>>>>
>>>>> One comment about the content: I would prefer to have a front page
>>>>> that just states what this thing is, and the projects on a separate page. I
>>>>> would like for it to be *super* easy to add projects. The whole idea is
>>>>> that the NIPY community is not a static list of projects, but an
>>>>> ever-evolving ecosystem.
>>>>>
>>>>> Cheers,
>>>>>
>>>>> Ariel
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>> Best,
>>>>>>
>>>>>> Vanessa
>>>>>>
>>>>>>
>>>>>> --
>>>>>> Vanessa Villamia Sochat
>>>>>> Stanford University
>>>>>> (603) 321-0676
>>>>>>
>>>>>> _______________________________________________
>>>>>> Neuroimaging mailing list
>>>>>> Neuroimaging at python.org
>>>>>> https://mail.python.org/mailman/listinfo/neuroimaging
>>>>>>
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Neuroimaging mailing list
>>>>> Neuroimaging at python.org
>>>>> https://mail.python.org/mailman/listinfo/neuroimaging
>>>>>
>>>>>
>>>>
>>>>
>>>> --
>>>> Vanessa Villamia Sochat
>>>> Stanford University
>>>> (603) 321-0676
>>>>
>>>> _______________________________________________
>>>> Neuroimaging mailing list
>>>> Neuroimaging at python.org
>>>> https://mail.python.org/mailman/listinfo/neuroimaging
>>>>
>>>>
>>>
>>> _______________________________________________
>>> Neuroimaging mailing list
>>> Neuroimaging at python.org
>>> https://mail.python.org/mailman/listinfo/neuroimaging
>>>
>>>
>>
>>
>> --
>> Vanessa Villamia Sochat
>> Stanford University
>> (603) 321-0676
>>
>> _______________________________________________
>> Neuroimaging mailing list
>> Neuroimaging at python.org
>> https://mail.python.org/mailman/listinfo/neuroimaging
>>
>>
>
> _______________________________________________
> Neuroimaging mailing list
> Neuroimaging at python.org
> https://mail.python.org/mailman/listinfo/neuroimaging
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/neuroimaging/attachments/20150822/0dfae680/attachment-0001.html>