linkeddata/dokieli

:bulb: dokieli is a clientside editor for decentralised article publishing, annotations and social interactions

dokieli's Issues

Improve page orphans and widows

Remove html5sortable dependency

This is a nice little feature which helps to reorder the table of contents, but then also helps reordering the sections in the document. It depends on jQuery and #47 .

UTF-8 characters and font support

Have a sensible way to use fallback fonts in which "most" of the Unicodes can be covered across OS/device.

Create / display chains of sequential or related documents

Examples:

• Paper v2 is derived from paper v1 + rebuttal + reviewer feedback
• Sequential chapters
• Paper is a follow-up from a proposal

These quite specific relations could be displayed differently from replies/annoations in the UI (from both ends of the relation, eg. derived-from and has-derivatives).

RDF lib for parsing and dependency

SimpleRDF and its dependencies e.g., https://github.com/rdf-ext/rdf-ext , https://github.com/alexmilowski/green-turtle, at the moment are too bulky in size. I think the only requirement right now is just RDF parsing and an internal graph to work with, must haves: Turtle, RDFa, JSON-LD. The rest of the RDF lib features e.g., LDP store, are not needed at the moment.

Note taking

Implement a note taking feature. Suggested by @clange . See also http://en.wikipedia.org/wiki/Org-mode

Footnote support

0a. "Having to read footnotes resembles having to go downstairs to answer the door while in the midst of making love." — Noël Coward

0b. Footnotes, endnotes, bibliography, references are all citations, and their distinction is not always clear.

At this time, browsers do not have very good (or at all) CSS 3 footnote support. If and when it is possible, create the styles and possible interactions for it.

Refactor UI editing features

Refactor the UI components which have to do with editing/authoring. They should only be available if the editing is enabled e.g., local storage, ToC sorting, data embedding.

Use ARIA

Remove the !important rules from CSS

There shouldn't be a need for !important in most cases. If do.css is using them, revisit to see whether it can be solved in other ways. Perhaps improving the existing stylesheets so that do.css doesn't have to use them.

New docs are created silently

Upon create, redirect to new document, or open in a new tab, or provide a link to click to open it.

Automatically generate inter referencing

There are internal links to the objects e.g., figures, tables, references, which are currently entered by hand 1) at the time of declaring the citation/reference point and 2) at the time of describing such reference. To simplify this process for the author and to reduce errors, investigate how this can be best achieved e.g., declaration of a reference point auto-generates a spot for it later in the document.

See also: #8 (comment)

better handling of narrow windows

It is good that the text display does not widen for wide windows, but it should be possible to have the text reflowed for narrow windows. This probably should not be the default option, but there may be situations where narrow display of the text is warranted.

Misplaces fragment border in right column hovers in ACM

When the mouse hovers on the containers in the right column of the ACM stylesheet, which have a fragment available, it shows the left border of that fragment block only on the left column. The expectation is that, hovering the left column will show the border on the left of that column, and hovering on the right column shows the border immediately to its left i.e., sort of in between left and right column.

Check and manipulate rendering of elements in different styles

In order to view the same HTML in different styles e.g., ACM, LNCS, the order in which some components appear in the document should be double-checked . For example, the way ACM and LNCS displays the figure captions are reversed in relation to the figure. To show the proper style for an HTML with different order (e.g., figcaption appearing before or after the figure object), either CSS or JavaScript should be used to position the components into their expected locations.

Remove jQuery dependency

Should work with vanilla JavaScript.

Ping a container/endpoint on document update

Feature proposed by Jens Wonke-Stehle:

"Ping a library/archive whenever a version of a research webdocument is saved" - https://twitter.com/wonkestehle/status/669187970168156160

"when you hit 'publish' it uses a to be specified way to notify a certain institution" - https://twitter.com/wonkestehle/status/669196303516180481

Eliminate any HTML differences for LNCS and ACM styles

If there are any differences in HTML for LNCS and ACM examples, make sure that they are removed. If possible, there should be no dependency from the HTML i.e., a single HTML should be used, while adhereing to the LNCS and ACM styles.

Edit menu too wide

When the edit menu is opened too close to the left or right of the page, the part that falls off the side of the page can't be accessed. Also it appears that it's not always appearing directly centred over the selection (not that this would resolve it if the selection is close to the edge of the page).

Make it so

Keyboard shortcuts

Make keyboard shortcuts for everything on the right side menu, under Do in particular.

Default to XHTML5 exporting

Currently, the Blob from the DOM is saved as HTML5.

The default exporting of the HTML should be well-formed to XHTML5.

Nanopublication UI

Create a UI to read/write nanopublications.

Trigger git commits on document updates

Eg. if editing a document locally, a script could run that automatically commits to a repo when the document is saved, and/or can be manually triggered from the UI.

Create error, warning, success messages where appropriate for user actions

Create how-to documents

Some how-to documentation would be nice both for developers and non-developers. What to expect and not..

It should cover topics/steps like:

• How to get a WebID
• How to have/use a personal online datastore (pod)
• How to format/create different types of references in a document
• How to annotate a document
• How to share items in a document
• How to manage documents
• How to use the menu operations (eg., sign-in, save as)
• How to version documents and interlink

Some of this information can be integrated into issue #36

Clickable links in PDF output

If PDF output is generated, make sure that links can be clickable. Suggested by @soeren1611

Consider replacing Computer Modern fonts

While the CM Unicode fonts are "pretty", it is only used for the LNCS stylesheet. It'd be better to drop them off to make the repository lighter as well as the page load when LNCS is default.

Revisit the whole CM support in systems. There is "CM Serif" available in Ubuntu I think..

Ensure that examples don't have horrid display

https://github.com/csarven/linked-research/blob/master/linked-sdmx-data.pdf
looks horrid. Two floats overflow their boundaries. Several displays use low resolution graphics. One colum is mostly empty.

Anyone looking at this example would conclude that the system is fatally flawed.

WebKit print issue with columns

Columns (in ACM style) in print are not applied in WebKit. See open issue (since 2012): https://code.google.com/p/chromium/issues/detail?id=99358 . Documenting this for the future.

floating figures

It would be nice to have figures float, i.e., appear beside the text, and move along with the text where the figure is being discussed.

Add a Table of Abbreviations

See original pull request for this feature and discussion #31.

Table of Abbreviations (i.e., using abbr and/or dfn information) should extend LR.U.buildTableOfStuff.

• add abbr option to tableList (and respectively update the switch). You would use the <dl class="toc"><dt><dd><dt><dd>.. for this - handle/assign that in the switch as well (i.e., ol and dl patterns) Everything else should remain the same. May need to update the stylesheets e.g., is there a default layout for LNCS, ACM etc for listing abbreviations/terms? If not, apply styles which blends with or follows the style guideline of each layout.
• do not onload when DOM ready because it is not necessarily a preferred feature by default. It should be manually triggered through the UI; in fact, same type of behaviour goes for the rest of the types of tables - this feature should be a different commit. It may be the case that some layouts will prefer to autoload all table of stuff e.g., thesis, which means that each layout should trigger whatever it wants to itself.

Handle table flow with many columns

When there are too many table columns, it doesn't fit into the page layout's column i.e., overflows out of the grid. There should be a way to handle this "quietly" without the author involved so that the table is displayed nicely. Perhaps taking up both page columns.

Related to issue #3

popups for links

Links to references should pop up the reference, as in Wikipedia, and not take you to the references section.

Clicking on a link where the target is already visible should not move in the document, but should instead highlight the target.

less aggressive links to figures

When you click on a link to a figure, that figure is moved to the top of the window, generally putting the link out of sight, even if the figure was already visible.

It should be possible to have a link to a figure pop up the figure instead of jumping to the figure.

What formats do Springer & ACM accept?

What formats do Springer & ACM accept? (LaTeX, Word,...)
Social hacking: get publishers to accept HTML version

Section headings 3 and 4, and text justification

LNCS author guidelines specifies 3rd and 4th level headings to follow with text which is to be justified. At the moment, the CSS doesn't justify the text (e.g., paragraph) which follows these headings (e.g., h4 and h5).

Create a dokieli homepage/demo document for reference

This document will be used to:

1. explain dokieli's technology and design principles
2. at the same time demo itself as to what's possible
3. and also purposed for doing tests

Using dokieli to read and write documents

It's not immediately clear (at least to me) how to use dokieli to read and write documents either from the README or from the website.

Could a short guide be written.

Experiment with OpenType features

Right now this only applies to LNCS as we are using CMU OTF.

Tests should be done across different rendering engines.

Tests should also try to catch any performance issues across devices.

At the moment, we are going with text-rendering:optimizeLegibility. It appears to be enabled in Gecko, but WebKit (and perhaps others) doesn't have it enabled. They default to optimizeSpeed.

Reinvestigate the use of Helvetica

Helvetica has highest priority for ACM SIG template's titles, however, if a system doesn't have Helvetica, printing to a PDF breaks rendering. Even though printing to paper is okay. Currently using "Times New Roman" for the purpose of not breaking. May need to be tested again as this stylesheet was developed awhile back.

Explore the generation of PDF from HTML

The simplest approach (for most readers) is to print the output to PDF from the Web browser. (Aside: that may or may not be a use-case, but we put that aside for now). However, this currently comes with inconsistent rendering due to different browser engines. Alternative ways to generate the PDF would be using a command-line transformation from HTML.

• Direct transformations from HTML to PDF should be tested
• Transformation from HTML->(La)TeX->PDF should be tested (as the engine and tooling behind (La)TeX to PDF may be more appropriate). Suggested by @kjetilk

UI for RDFa embedding in prose

testing on math-heavy documents

Mathematics is a problem for HTML, even with MathML.

There should be comprehensive testing on math-heavy documents.

better control of figures needed

It should be possible to have a figure appear in one place in the displayed output and and another place in the printed output.

In displayed output the figure should probably appear next to the first reference. In printed output the figure is often best placed at the top of a page or column, before the first reference.

self-contained documents

It should be possible to bundle up an entire document into a single file that can then be displayed offline.

Creating a zip file for the document is not adequate, unless there is a simple way to display the zip file, i.e., without first having to unzip.

Ideally all one would have to say would be open <filename> to display a document in one's browser.

bad links in HTML output

In http://linked-research.270a.info/ the link for Figure 2, [2] is
http://linked-research.270a.info/#figure_lsd-sense-worldbank-2012-topic-4
but that does not appear to be a valid link in the document.

Alternative authentication

eg. Something with bearer tokens so I can authenticate with (and potentially post to) my own website (that isn't running solid) from dokieli.

Logo

dokieli needs a logo.

• logo
• favicon

Dynamic page breaks and content flows

There is some basic page breaking handling but some improvements or more tests would be nice. While the author can always manually adjust where the page breaks, it would be good to handle better automatically with CSS and/or JavaScript.

Similarly, it would also be good to appropriately readjust the order of the objects on the page.