The MediaWiki:Site improvements page is a catalog page for implementations, extensions, upgrades done or in need of doing to the coding of the two MediaWiki sites Hmolpedia (Hmolpedia.com) and Hmolpedia 2020 (EoHT.info).
In Sep 2020, the MediaWiki.org software platform, via the server host A2Hosting.com, became the new “wiki” framework for Hmolpedia, previous hosted via the WetPaint.com wiki (2007-2013) turned WikiFoundry.com (2013-2020) wiki platform. During this transition, the encyclopedia was split into two versions, an archived version (EoHT.info) and a newly-written edition (Hmolpedia.com), wherein the extant 5,376 articles, penned via the old wiki platform, as of the end of Aug 2020, where placed as html files into the MediaWiki-based EoHT.info URL, renamed as Hmolpedia 2020.
The following are issues open and in need of solving, ordered via of rank of pressing importance (top being in most need of fixing):
Math parsing | Issues
The following math parsing errors occurred (12 Jan 2021) on the Gibbs energy article:
This occurs frequently, when a growing number of math or chemical equations are added into paragraphs?
5,376 Article Migration | Viewable Online | Status: OPEN
In 2 Sep 2020, the 5,376 articles of Hmolpedia, in the WikiFoundry servers, were extracted into a zipfile of html articles, in a “pages” folder, indexed alphabetically (0001, 0002, 0003, etc.) as shown below:
The following, e.g., is file #0002,
placed in the A2Hosting File Manager, location:
a screenshot of which is shown below:
Which shows the “-ism” shortened (without the indexing “0002%20”) and in the original indexing form:
A. as MediaWiki articles
- A. Ideally, we would like these 5,376 html files to be, somehow, placed or converted into MediaWiki pages, so that all the hyperlinks “work”. Presently, e.g., if we click on the “eliminative materialism” article, from this new-viewable (as html page) article:
we get the following uncreated wiki page as follows:
Nischay has communicated, as far as I can tell, that
- “In MediaWiki, pages are not stored as HTML files. Instead they are saved in the database in wikitext format and only render as HTML.”
- — Nischay Nahata (2020), “Email to Libb Thims”, Sep 14
Hence, if I am understanding things, there is no way to convert the 5,376 html files into a database of wikitext?
HTML to Wiki markup / syntax
The following links seem to be able to facilitate the conversion of HTML pages or files to Wiki Syntax:
Note: In the film Social Network (pdf pgs. 13-16), the go through how to pull images off a website and collect them into a file.
B. HTML viewable
- B. Alternatively, if we cannot convert the 5,376 html files into wikitext (in the database), then option #2 would be to put all the 5,376 html files into a “pages” folder, but remove the auto-indexing (0001, 0002, etc.) and the html extension, so that the pages are viewable in html AND the links in the html articles “work” properly, e.g. clicking on the “eliminative materialism” link (in the -ism html version of the article) brings up the “eliminative materialism” article (no index and no html extension), as follows:
Such as currently seen here (in the WikiFoundry subdomain):
Notice that the plus sign “+” in the article file, does not show in the article title (of the wiki article), whereas it does seem to show (in the wiki article) of the MediaWiki version.
~/page/article to ~/wiki/article "folder" change (in the EoHT.info wiki)?
Note: in the EoHT.info wiki, if the 5,376 html articles are to be put into a “page” folder (so that the hyperlinks work), then the current “page” (~/page/article) will have to be stitched to “wiki” (~/wiki/article)?
On 24 Sep 2020, Libb suggested to Nischay the following video, which does on engine rewrite, on how to get rid of the html extension:
0001 HTML article indexing prefixes
On 25 Sep 2020, Nischay suggested that we use batch scripts to rename the 5,376 html files without the indexing:
Capital vs lower case file issue?
As of late Nov, the HTML versions of the 5,376 files seem to be working good, with the exception that all the article names are saved as first letter capital format, hence lower case searches and links, e.g. "chemistry" give 404 error (as compared to Chemistry, which works):
The following are some seemingly related reference articles I've found so far:
- Convert and redirect URL in upper to lowercase using .htaccess (2017) – ServerFault.com.
- Are URLS Case Sensitive? How to Redirect Uppercase URLs to Lowercase URLS Using Htaccess – GoInFlow.com.
- RewriteRule lowercase transform for file names only (2016) – StackOverflow.com.
Symbols | 404 error
Nischay got the “+” symbol, which was the previous term symbol joiner of the previous wiki file method, to redirect correct.
The “:” and the vertical bar “|” still, however, need fixing:
WikiHiero Textwrap | Status: OPEN
On 15 Sep 2020, Libb Thims installed the WikiHiero extension, and it worked good.
On 24 Sep 2020, following the MediaWiki 1.35 update, articles with WikiHeiro on them returned “fatal errors”, after which Thims updated the WikiHeiro extension to the 1.35 version, and now works good.
An ongoing issue, is that the text doesn’t wrap around each hieroglyphic like it is a "word", such as shown below for the Wallis Budge paragraph:
Such as it works in the WikiFoundry version of the Ra article:
On 25 Sep 2020, Nischay added a CSS page “MediaWiki:Common.css” page to Hmolpedia.com, and it made the text wrap around the WikiHeiro images, but messed up the image and equation rendering for the rest of the site (as discussed here).
The following are articles "in limbo", in an undecided state or condition, or somewhere between open and closed:
User permissions | Status: In LIMBO
Related to ConfirmEdit and QuestyCaptcha, eventually, when both sites come online, in respect to high traffic flow, people will attempt to aggressively edit articles to their own liking, delete things, etc., hence user permissions will need to be changed; to approximately the following:
- EoHT.info (only Libb Thims can edit)
- Hmolpedia.com (will leave as is for the moment, until more user data comes in)
Of note, in 2018, Hmolpedia, owing to a hacking spree (which removed 100s of articles), done by a new user, was changed from “anyone can edit” to writers who “applied to be writers” could edit. 
Extension:Cite | Status: Working (but maybe add other variants?)
On 23 Sep 2020, or shortly before, Libb Thims added the “cite” extension, which allows the "ref" and " ref name="name" " parser hooks to be used in articles. These work pretty good and do the job.
There seem to be other "citation" implementations that could be tried, e.g. this video has an auto-generate citation button in the Visual Editor.
Also, this video, has this option when the VisualEditor citation button is clicked:
Printing | Status: OPEN
A noted downside of Extension:Cite is that when articles are printed, the bloated code of each extension shows up and is printed within the paragraphs of the printed page, which makes the article nearly unreadable, as a printed reference article?
The following are issues closed and solved:
Table Row # Auto-index | Status: SOLVED (85%)
On 16 Sep 2020, Libb Thims communicated to Nischay that he has many large tables, with 100+ rows, e.g. the top 1000 geniuses, that need to have an auto-indexing feature, so that when one row is moved up or down, all the row #s reformat, automatically, such as attempted here:
In the Wikipedia List of oldest living people (50 names), they use the “template:row numbers”, but I read that this does not work in mobile.
On 23 Sep 2020, Ike Hecht suggested an approach where an auto index number is prepended to every row. See the example here: https://stackoverflow.com/questions/17012421/auto-number-table-rows/17012547#17012547
In Nov 2020, Nischay got the indexing feature to work for tables, on both desktop and mobile view, which start with #1 (e.g. geniuses references list); but for a list that starts at say #201 (e.g. geniuses 201-400), he says this will need a special rewrite rule?
Mobile Version | Status: SOLVED [or use better version?]
Google Search Console reports that Hmolpedia.com has “mobile usability” issues.
On 30 Sep 2020, Libb Thims installed the Extension:MobileFrontend, and tested it on his phone, and it seems to work (e.g. the phone loads in mobile version, and there is button for “desktop” at the bottom).
Will check back in Google Search Console, in a few days, to see if problem is resolved?
If someone knows of a better “mobile extension”, please advise.
Visual Editor | Status: SOLVED
On 21 Sep 2020, Nischay Nahata was able to upgraded the Hmolpedia.com MediaWiki (but not the EoHT.info MediaWiki) to MediaWiki 1.35 and got the Visual Editor to work.
Index.php vs page folder | Status: SOLVED
By 15 Sep 2020, Nischay had switched the “index.php” folder/file name of newly-generated articles, to “page”, so that new articles generate as follows:
instead of the old way:
Spam Bots | Status: SOLVED
On 14 Sep 2020, as shown in Special:Version, Thims added ConfirmEdit (email confirmation to join) and QuestyCaptcha (need to answer one of three hmolscience-related questions to do ANY editing). Stopped the spambots cold.
Math Equations | Status: SOLVED
On 15 Sep 2020, Libb Thims installed the SimipleMathJax extension, but it had issues, namely (a) the equations rendered “center” (which was undesirable), and (b) the equations were added in a delayed "ghost-like" fashion, compared to the speed of page loads; i.e. had to go up to the cloud, then back down into the page, before they were visible, which made for a very unprofessional (time-delayed) look to articles.
On 23 Sep 2020, Nischay explained to Thims how to “update script” in the file manager, after which Thims was able to successfully install the standard “Math extension”, which works good (right centers) in Visual Editor.
Short URLS | Status: SOLVED
By 15 Sep 2020, Nischay had fixed the URLS of both sites, so that the the main pages display as follows:
Instead of the original setup version:
Chem | Parser hook
A noted core parser functions, and or one that comes with the "math" extension [?] is that the one can render chemical formulas, such as H20 or H2SO4, using "chem" parser hooks, in edit source mode, as follows:
which seems to be quicker than the H_2O method, e.g. one can render CO2 inline as (with the chem parser hook, which compares to CO2 (with VisualEditor); see the MediaWiki "parser extensions tags" article, for a list of extensions available.
Chemical element | Parser hook
Another noted extension is the "ce" parser hook which makes chemical elements look more refined and bigger in text, for example:
- The element helium (with ce parser hook) vs He (no hook).
Todo: Possibly add the Extension:YouTube?
- Progress report (WikiFoundry subdomain) – EoHT.info.