Update website #27
Comments
|
Updates, largely taken from email convo with @DrDrij: re: item #1 - Andrij has created a 'dummy' page to showcase how we could display our content. It's here at http://beta.econ-ark.org/demark/. @llorracc, @mnwhite, what do you think? If it's not what you imagined, can you try to articulate what's different/missing? re: item #3 - we've got a first draft set up at https://econ-ark.org/acknowledging/ - @llorracc if you could take a look and give feedback? re: item #5 - @DrDrij, probably the simplest thing to do would be to display a Google calendar, either embedded or through using an API/plugin to get the information and show it on the page. |
|
re: item 1: I just took a look at the "dummy" page. It looks pretty reasonable, except that I'm assuming that the unattractive gray color reflects somehow the fact that it is a beta/demo project, and is not the proposed actual color scheme? (I'm guessing this partly because even the Econ-ARK logo is gray). But I'd like to make sure we explore what other projects have done for their content management and indexing -- especially QuantEcon, AstroPy, and other NumFocus projects. Maybe this is something to ask of the NumFocus "contact group" -- are other people using some kind of content management system, and if so, what is it? But I'm MORE concerned with the issue of how we construct a lightweight and low-maintenance system for organizing our content, than I am in the appearance of the display. I have a strong sense that the right way to do this is via metadata/tags embedded in the notebooks themselves, so that there is a "single source of truth" and there's no way to get multiple conflicting versions of things like which tags go on which notebooks, etc. For #3 -- I love it, perfect, exactly what I wanted (except that maybe we could have the little tool for copying the links to your clipboard, like mybinder.org has when they show their badge?) For #5 -- we already have some Google calendars set up on the [email protected] page, but they have not been used for much (and are far from being complete). We should probably clean them up a bit and then use them as the source for whatever we do here. Something else we might want to think about is ways to automate connections between our github presence and a calendar; for example a quick Google search led me to this https://zapier.com/apps/github/integrations/google-calendar which says that you can have the calendar auto-populate via a bot whenever, for example, we mark a new "milestone" has been achieved. |
|
Thanks @shaunagm for the write up, @llorracc for the feedback. 3 - Acknowledging I've updated the page with HTML and Markdown snippets for users to copy the badge source easily. As per your suggestion, we've also done the same page over at QuantEcon.
Would you be interested in me creating a similar logo for Econ-ARK to replace the badge? 5 - Calendars @llorracc @shaunagm Is it possible to set the [email protected] calendar to public? You'll see on the test page below I've successfully embedded the Google Calendar but there are access permissions, if you aren't signed in. (I'm having issue logging into the Econ-ARK google account from Ukraine) |
|
Andrij,
I've just created a public calendar which should be accessible at the link
below.
Anything I can do to make it easier for you to log in from Ukraine?
https://calendar.google.com/calendar?cid=amh1ZWNvbi5vcmdfdDMzN2Q1czF2Z2hiaWV0bDBrcHA5YWwyNGtAZ3JvdXAuY2FsZW5kYXIuZ29vZ2xlLmNvbQ
…On Tue, May 21, 2019 at 12:48 PM DrDrij ***@***.***> wrote:
*4 - Donate*
Thanks @shaunagm <https://github.com/shaunagm> for the NumFocus link.
@llorracc <https://github.com/llorracc> Is there anywhere in particular
on the homepage you'd like the donate button?
- Button style
- Button location
- Copy to go around the button
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#27?email_source=notifications&email_token=AAKCK72Q7R74P4KCTXDMPSTPWQRWXA5CNFSM4HHPVQP2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODV4QKQY#issuecomment-494470467>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAKCK7ZW3NJ5EKVRDIP2JPTPWQRWXANCNFSM4HHPVQPQ>
.
--
- Chris Carroll
|
|
@lisamartin72 see #4 above |
|
@DrDrij - we've talked some more about generating the metadata for item 1 above. Are we going to be getting our notebooks by using the Github API and if so, can we also look for a metadata text file in the same directory in order to generate tags? How's that for a solution to this question? |
|
To elaborate: We came up with the idea of incorporating our metadata (tags,
etc) in a structured companion file, so like the jupyter notebook
BufferStockTheory.ipynb would be associated with BufferStockTheory.json
which would contain the metadata. Or maybe BufferStockTheory.rst, if we
want to use RestructuredText as QuantEcon is doing. There must be tools
for parsing either json or rst files to extract their content.
The process of creating a REMARK or DemARK or other resource could be
structured through the website, by answering some questions which are used
to generate the .rst or .json file, but then the files themselves could be
updated easily and on the fly with normal GitHub procedures.
…On Wed, Jun 5, 2019 at 10:51 AM Shauna ***@***.***> wrote:
@DrDrij <https://github.com/DrDrij> - we've talked some more about
generating the metadata for item 1 above. Are we going to be getting our
notebooks by using the Github API and if so, can we also look for a
metadata text file in the same directory in order to generate tags? How's
that for a solution to this question?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#27?email_source=notifications&email_token=AAKCK75MBMUPXSOZXCILW33PY7HIRA5CNFSM4HHPVQP2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODW762HA#issuecomment-499117340>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAKCK73LPIUONDPBCY47QW3PY7HIRANCNFSM4HHPVQPQ>
.
--
- Chris Carroll
|
|
I've given some thought to how to serve up our notebooks to users via the website, which has been shaped by my efforts to compare MyBinder, Google CoLab, Microsoft Azure, and other options. Basically, it is not clear which of these options we might want to use as our default -- and, in fact, the answer might be different for different notebooks, or it might be systematically different for DemARKs and REMARKs. So far my approach has been to put clickable buttons at the beginning of a notebook with a link that will fire up either CoLab or MyBinder. But that's unsatisfactory for several reasons:
PS. We could contemplate that part of the metadata associated with each notebook should be a suggested choice about how to serve it up to users (binder, CoLab, Azure, etc). |
|
@llorracc I've only recently been sent a brief on a new series of lectures under QuantEcon. Their proposal is to have a "dropdown" element allowing the user to launch the notebook in any respective platform. Noting: an equivalent element would live on the econ-ark.org website next to a notebook on the listing page. However, this is again going against adding such metadata to a notebook or adjacent file. I don't believe this data should live alongside the notebook itself. |
|
@llorracc I've only recently been sent a brief on a new series of
lectures under QuantEcon. Their proposal is to have a "dropdown" element
allowing the user to launch the notebook in any respective platform.
Excellent! Having spent some time in the last month exploring the options,
I had been thinking about asking you to set things up so that on
econ-ark.org we could switch between different hosting methods.
I think, though, that maybe the best way to do it is to have a metadata
element that specifies either a "preferred" or a "required" environment for
each notebook. If no such element is specified, then there would be a
default. The reason to do this is that some notebooks may be better suited
to one platform than another. The leading example would be notebooks that
require the full capabilities of LaTeX rather than the subset that is
available by default.
This seems like a good goad to us to think about ways to have our approach
to notebooks converge as much as possible with that of QuantEcon. (And
maybe vice versa).
We have both struggled to reconcile the procedures of GitHub with the
structure of Jupyter notebooks, but QuantEcon has decided to build the
Jupynx tool to allow the notebooks to be written in rst, while we ave (so
far) adopted the more lightweight approach of using the jupytext plugin.
The advantage of the jupytext approach is that you can do all of your edits
in the straight jupyter notebook without having to translate anything.
Maybe there could ultimately be a fusion in which draft and early versions
of notebooks get written using the regular technology, and as a package
matures it migrates in the direction of the more rigid rst format.
I'm quite keen to integrate whatever we do with whatever QuantEcon does,
and am willing to compromise as necessary to get there.
…On Mon, Jun 10, 2019 at 7:25 PM DrDrij ***@***.***> wrote:
@llorracc <https://github.com/llorracc> I've only recently been sent a
brief on a new series of lectures under QuantEcon. Their proposal is to
have a "dropdown" element allowing the user to launch the notebook in any
respective platform.
Noting: an equivalent element would live on the econ-ark.org website next
to a notebook on the listing page.
However, this is again going against adding such metadata to a notebook or
adjacent file. I don't believe this data should live alongside the notebook
itself.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#27?email_source=notifications&email_token=AAKCK7ZSCRNIOYYMF6HR44LPZ2E67A5CNFSM4HHPVQP2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODXKRMDA#issuecomment-500504076>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAKCK75YAG6DCAWZ677BBIDPZ2E67ANCNFSM4HHPVQPQ>
.
--
- Chris Carroll
|
|
Item #3 - Added a link to acknowledging Econ-ARK (https://econ-ark.org/) Item #1 - Keen to move forward with this. @shaunagm Would you be comfortable working together to create a series of scripts, leveraging Github API to generate notebook listing page for the site? Further to this reading through a metadata companion file for additional information display. |
|
For Item 2, there's three wiki pages that may be relevant to the website. The first, for the econ ark team, should probably just be a "team" page on the website itself, similar to QuantEcon's team page with photos and everything. Once that's up, I'll delete the wiki page. Second is the current roadmap. I'd link to that with something like, "learn more about our plans for Econ-Ark". @DrDrij, please let me know before that goes live (or at least immediately once it goes live) because if we're linking directly to it for newcomers I probably want to provide a paragraph of introduction about how the page is structured. Finally, there's the open source economics software page. This is new and a work in progress, but the overall goal is to be a place to collect information about the community. Perhaps it could be linked to with the phrase, "The Econ-Ark team also helps organize the community of open source economics software developers." I'll let you know once the new docs site is up and help you create links to that. |
|
Oh and re: item 1, Andrij and I decided on the call that we're going to go with the simpler solution of having a CMS that links to notebooks, with the metadata for the notebooks stored in the CMS. Once this is done, we'll give it some time - a month or two, perhaps - and if the lack of a single source of truth is a problem we'll then implement the more complicated option. Basically, let's make sure we need the more complicated system before we go through the work of implementing it. |
|
Moving this from email: the page for the events calendar: https://econ-ark.org/calendar/ |
|
@DrDrij - in anticipation of our call next week, can you take a quick look at the original post of this issue and let me know if it reflects the current status of the site? Are there things you're done with that are not checked off? Are there things that you're working on that are not included in the list? |
|
Thanks @shaunagm. I will work them into the new site! |
|
@shaunagm Item 5 and item 8 can be checked. They've been worked into the redesign. For item 2 - I propose we decide on where these links should live whilst creating/updating final copy for the home and about pages. |
|
The "How to Cite HARK" page looks great, but I found that there wasn't a cut-and-paste option that was appropriate for slides for someone who used HARK in producing their results. Just putting the full Econ-ARK logo at the bottom of the intro slide makes it look like the project is a part of Econ-ARK, rather than just having used Econ-ARK. The "Powered By Econ-ARK" badge would work except that the code to use it is html rather than LaTeX. PS. Also, we need to figure out what to do about #309, about using the CFF format, which seems to be an emerging standard. |
|
Hi @DrDrij, how are things going? Also: is there somewhere we can stick a link to this podcast episode about EconArk? |
|
Checking in on progress on notebook launching, and with an update. @MridulS has somehow managed to set up a scheme (you'd have to ask him how he did it) that launches our slowest-to-launch notebooks in a few seconds rather than 3-5 minutes. If what he has done is sustainable, we should at a minimum include it in the set of options for launching notebooks. (somewhat) relatedly: We've had a number of inconclusive conversations about how to deal with the fact that different notebooks have different requirements. It's surprising to me that there's not already a standard for this (probably there are worthy persons debating with each other about how to do it). But I'd propose a two part solution. We should define a core set of requirements that every notebook can expect, which would include numpy, scipy, matplotlib, and a few other things. Then we can define some technology to specify notebook-specific requirements in addition to the baseline requirements. Maybe there would be a "master" version of a notebook that would pip install everything, but if the notebook was launched with (say) Mridul's tool via econ-ark.org, the version that was launched could strip out the "requirements" cell. Basically, my goal is to have an infrastructure where we can send a command "launch the KrusellSmith notebook" and the back end makes decisions about how to do that (so that, if we change our minds, we can change the back end). |
Regarding the different requirements and initial setup for notebooks, I am fiddling with the idea of creating an Econ-ark specific docker container, which will create a base environment for all the notebooks in the econ-ark organisation, this will include the standard requirements like numpy, Matplotlib, .. and latex. We host the docker container and instead of fetching the minimal python docker container, our jupyterhub/binder will fetch this custom docker container. This workflow can be an issue if we want to stick specific version of libraries to notebooks. For example notebook_1 works with numpy v1.14 but will break for some future version of numpy and we want to move to the current version of numpy in our econ-ark docker container. While writing this out, I think something like [(minimal python + latex + Jupyter_extensions) docker container + requirements.txt] for python requirements should be able to solve the problem I just mentioned. (probability of backward incompatible changes in latex is extremely small) |
We can create forwards from econ-ark (from the buttons) to specific notebooks hosted on our deployment of binder. Like clicking on bufferstocktheory button on the new website will forward the user to http://bufferstock.computedata.science |
I've placed temporary update message on the Econ-ARK homepage. Please let me know if you'd like it adjusted (or removed).
Other updates around the new site at http://beta.econ-ark.org/
@llorracc What are your thoughts on this new beta site? Shall we go ahead and start populating it?
Thats amazing! Nice work @MridulS. A question around this - are we interested in users having the option of launching on different platforms or focus on the one we suggest and configure? |
This issue contains a list of items we'd like to update the website to include.
Major
1: Create or find a system for managing metadata of our HARK user content. We've got four broad categories of content: demonstrations (jupyter notebooks demonstrating usage of HARK), replications (jupyter notebooks implementing existing papers in HARK), instructional tools (notebooks constructed for teaching purposes, including problem sets) and presentations ( notebooks and non-notebook content from presentations given related to HARK/econ-ark). These will be stored in different folders of a single Github repository. Ideally this content metadata system will use tags.
2: Work with Shauna to add links to relevant off-site content such as the new docs site and the project wiki.
3: Add a "how to cite HARK" page.
Minor
4: Add this donation link somewhere on the landing page: https://numfocus.salsalabs.org/donate-to-econ-ark/index.html
5: Create an events calendar/embed a google events calendar (there's a small amount of possibly not useful stuff in private google drive in web/move_to_website/events, Shauna can get this for you)
Ideas we've not committed to yet
6: Create email addresses with the econ-ark.org domain name.
7: Add a blog to econ-ark.org/blog. Not sure yet what will go there and therefore what kind of blog to use (Wordpress, something hosted and searchable by Github, hosted by QuantEcon, etc).
*More stuff
8: Rethinking calls to action.
9: Add CFF citation
The text was updated successfully, but these errors were encountered: