vanch3d / nvl-slim Goto Github PK

Need to implement Twig Extensions (or something else) to replace context-dependent formatting of customs items (github & linkedin widgets in particular)

See https://www.crossref.org/display-guidelines/

no more to say here...

The UI for images, both inline and on the strip, is confusing and, at the moment, clearly broken or badly configured. Need to rethink their integration

The default route to a publication directly leads to PubReader:

• awkward access for publication without plain text or PDF (e.g. source code, non-open copyright)
• no placeholder for overview and alternatives (e.g. full metadata, links to export and alternative sources, experiemtnal visualisation such as analytics and cloud, ...)

• Move all publication PDFs to Zonedo
• Check their API for back-end integration

Generating the data for file-caching at first request takes too long and might time-out.
It might be better to switch the data collection to an AJAX request on client side.
Or get a cron job to update the cache on regular basis.

At the moment, only the year seems to be retrieved but the structure remains CSL-compliant (I think), resulting in an obscure data package, e.g. "issued": { "date-parts" : [["2008"]] }
Need to look at the whole chain

Both /unapi and /unapi?id={pub}&format={format} with correct parameters are considered successful requests and currently respond with a 200 status.
The last version of the unAPI documentation has a cryptic comment:

302 Found for responses to the UNAPI?id=IDENTIFIER&format=FORMAT function which redirect
https://web.archive.org/web/20070107163855/http://unapi.info/specs/unapi-version-1.html

but no mention of redirection in the description of the body.
However, revision 1 of the documentation clearly states:

3. UNAPI?uri=URI&format=FORMAT
[...]
Redirects the user to the object specified by URI in the format specified by FORMAT. The content-type should be the content-type specified in the element within the unAPI format response for this format.
https://web.archive.org/web/20070102033638/http://unapi.info:80/specs/unapi-revision-1.html

/unapi?id={pub}&format={format} should be changed to 302

• validate all status codes
• fix the API handler
• fix Swagger accordingly
• fix all <abbr> identifiers in the templates

problem with d3.js?

For display in a continuous timeline, the scaling is ambiguous with broad dates.
Make sure to have at least year and month in the fields.

The "all publication" UI needs to be redesigned

For example:

• If the PDF of a publication is missing (/publications/2016.IMS.Inteliigent.pdf), return 500 (should be 404)
• if a publication id does not exist with PubReader (/publications/THISISAFAKEID/), return 500 (should be 404 or 410)

The SVG icon for the search button is not scaled down properly (on Firefox 57) or not displayed at all (on Chrome 63)

Note: switch to FA?

• LESS (or SASS?)
• bower
• Grunt

• Narrative and Network graphs to have similar elements such as the popup doc panel
• consistent responsive layout (Narrative scrolling & max size computation quite good)
• maybe using a reusable charting lib like c3

The main menu is static, including the project links.
Would be better to get it dynamically, maybe based on a "starred" attribute for the project

PubReader is doing a good job but is very old now and need either an overhaul or replacement by something more up-to-date.

To consider:

• publications need to be moved to a more suitable format (see JATS, https://jats.nlm.nih.gov/archiving/)
• coonsider testing browser or editor such as
  • Substance (http://substance.io/)
  • Achivist (https://github.com/archivist/archivist)
  • eLifeScience's Lens (https://github.com/elifesciences/lens)

• missing "title" attributes
• headings
• ARIA

Run checks and validation with WAVE

A great oversight: never commited the tests!
And, apart from the refactored unAPI ones, they are likely gone.
Check in other backups (if lucky) or recreate them as we go through more updates