User:Retro/Mental accretion
| Broad editing organization | More specific projects | ||||||
|---|---|---|---|---|---|---|---|
| Article improvements | Mental accretion | Phrases to watch | Templates | Character encoding | DKC organizing | Footnotes | Tooltips |
| If my subpages inspire you to carry out an idea yourself or you have any questions or concerns, I'm interested in hearing about it! | |||||||
Immediately pursuing my thoughts about how change Wikipedia is not always productive for my time or priorities. By noting things here, I escape my fragile human memory.
*Meta organizing*
[edit]- Might want to have a subsection for templates to create, vs. changing existing templates.
- Top-level sections (aside from this one and "...Miscellaneous") should be sorted alphabetically because it's more navigationally practical. If the sections are culled down, it may become practical to prioritize by impact on articles, but there's currently too many separate categories.
Automation
[edit]This is a combination of user script ideas, bot ideas, and general off-wiki script ideas.
Linking to diffs
[edit]I need to straighten out my linking of diffs. I like plainlinks for diffs, but it's not strictly necessary; I would just like to have the most efficient way. There must be a user script to create a wiki script editable version from the link on a diff page. If it does not already exist, I can create one myself easily enough by modifying my personal JS/CSS. I will want to look at existing scripts and script organization efforts first (I'm already familiar with popular ones, like Twinkle, Uchuka's cite refs (niche popular), and Huggle.)
This is actually useful to the Wikipedia general community that I think linking should be integrated into the diff itself. But those developing will probably say it's not worth developing because one can just copy the link anyway (which misses the point of plain-linking wiki-internal things).
And then there's Special:Diff, which works, but is a pain to use. I would much prefer to have title text that describes the diff (like Revision of PAGE_NAME at DATE), perhaps with conformance to existing phrasings of diffs if available. If I made my own script, this is a benefit that I would include, and possible look into including it in the original diff template. I wonder if subst: is good; it seems like switching to subst only for diffs would be beneficial, but actually, I don't personally care or understand the rationale for when templates are substituted versus when they aren't; it seems like a "madness of the crowds" random situation, where templates are randomly selected to serve this purpose.
So basically, it is a problem, but much more solvable than I originally thought.
- I can even use diffs as notes -- they can show an issue or a future change I want to make (sometimes I not future changes in edit summaries; probably an unwise practice for the sake of ever getting back to a change.)
- Disable enter key to save edits. I don't like it.
- Having Javascript that formats links to sections nicely would be nice to have. It should be relatively easy to automate with the section link template. It wastes so much of my time.
- There could be two variations, nicely linked and piped link.
- And also a permalink.
- I also might like a GUI button to mark something as code and surround with nowiki (or perhaps not).
If I'm going to switch to Special:Diff-style links using a userscript, I'm going to want to include the page title in a comment, since the diff itself could be rev-deleted, and the page itself could still be potentially useful.
I made a Python script for linking a list of diffs given a list of diff ids. It's not particularly robust, because it requires a list of diff ids to start with; it can't derive a diff from a diff url.
def num_link_diffs(diff_ids):
return ', '.join(f'[[Special:Diff/{id}|{num}]]' for num, id in enumerate(diff_ids, 1))
# A convienent shortcut for command-line use.
# The full name should be used when it's in a module (see also PEP 8).
nld = num_link_diffs
Scheduling notices
[edit]Schedule notices to myself. This will require clever template construction. There are 3 components that entered my mind:
- A date based filter, so the notice appears after a certain date (that's the point)
- A user based filter, so only I can see it. There are two ways of doing it:
- My original idea was to insert some template satisfying this criteria in the top of my user page. This has the advantage of being performance efficient, because non talk page locality is not.
- CSS-based implementation to appear on *all* pages (or even just my user page). If restricted through an intermediary template, it still seems like it might be difficult to implement efficiently (how would the display work anyway? there still needs to be a content-based list somewhere, so page editing will probably be inevitable anyway.)
- The same CSS format as a talk page notice. I've seen this done on MegaTron1's DKWiki page.
The utility of such a system is questionable... I haven't determined any situations where I've actually needed this, but if I do end up needing this, I will investigate it further.
The thing I have in mind when drafting this is the Dissident97 discussion, where he eventually nominated the list of video game soundtracks considered the best for deletion. {{small|On a side note, even though I initially argued against it, I can accept the decision and appreciate the reasons for it happening.)
I don't think there's a magic word to check for the current user. But eh, I guess my preference is not what a generally preferable feature anyway (it's specialized use, it could potentially be abused in article space, etc.). I could do the same thing by using !important in Javascript, with <span>...</span> that normally hide the content surrounding my header.
Featured article review bot
[edit]This is more of an off wiki direction tool idea than a bot idea.
- Featured article review bot; suggests pages that perhaps could use review based on how small their featured article section is. By suggest, I don't necessarily mean post a public comment; perhaps the bot could be used on a personal basis. I can't necessarily see a way for the bot to post a public comment without being a bit disruptive.
- I wonder what the largest current Featured Article is. It seems like many large topics have been listed and delisted (e.g. Islam). Perhaps large sized articles could be flagged for review by the bot from my previous FAC review idea. Or actually, it isn't currently previous, it's right here: There should be a bot to automatically assess old featured articles for insufficient discussions that were leading up to the featured article.
Automation miscellanea
[edit]- I will want to look into this searching tweak later.
- Wow, I *really* need the ability to annotate Wikipedia pages. I suppose I could continue my external noting format, or even have some semi-automated saving system built (a derivative of this idea), but I want convienance in noting page errors I'm unsure how to immediately fix.
- I want a search help link next to my top search bar. Like the ? icon on Special:Search
- I wish links to redirects in searching would at least include a direct link to the redirect (I can probably program a user script to allow this.)
- I could write my own user script to minimize {{talk header}} to a show more box and the archive searching, since that's the main functionality I find the template useful for. I still will probably keep the rest of the template in a show more section, since I might occasionally want a quick policy link.
- Not exactly a user script, but a general script: Wikipedia page history analysis tools: One thing I would like to see noted is a note when an edit brings a Wikipedia page back to an identical prior state.
- I would like a local Mediawiki instance for my favorite pages, like Donkey Kong Country, so I can more effectively track changes offline. Then again, I may be able to do even better with my own customized diff tool.
- I really need a way to automatically sync my personal files with my wiki files more automatically. Unfortunately, a page export would probably require a cross-export from XML back into non-XML. Ideally, I could just abstract it all away with a local Mediawiki instance.
- I would appreciate an in-interface magic words and parser functions helper. Or maybe I should just set aside some time (or schedule practice) and memorize them all. Probably both.
- The Mediawiki page list limits have been changed from 1000 to 5000.
Project pages
[edit]Here, I use "Project" to denote pages in the Wikipedia and Help namespaces.
Policies and guidelines
[edit]This space intentionally left blank.[a]
Historical pages
[edit]- For archiving, this is a very useful page I've found that covers many options: Wikipedia:Template messages/Wikipedia namespace. I'm unsure how comprehensive it is, and it seems a bit outdated or gives undue weight to certain templates in some places, but it's been generally useful in allow me to find well-suited templates.
- Help:Categories/feedback is definitely historical; it should possible be marked with {{Historical}}
- I wonder if there are any other 'feedback' subpages.
- Wikipedia:Requests for adminship/BAG: should be marked as historical.
- Here, I'm starting to believe I was wrong. These are quite relevant.
- Wikipedia:Version 0.7, Wikipedia:Version 0.8: These are definitely no longer relevant, perhaps worthy of {{deprecated}}, but may even be better with tailored deprecation messages. There are some other interesting similar pages that can be found looking through AllPages in the Wikipedia namespace, but it's probably largely a waste of time to do so.
Brilliant prose
[edit]I currently think Wikipedia:Brilliant prose should be restored from a redirect to a full page, with {{superceded}} on the top, and a few explanatory notes highlighting how the old Featured articles differed from the current FAC procedures. You can read my explanation below if you want, but it's still a work in progress.
Elaborating why Wikipedia:Brilliant prose should be restored.
|
|---|
|
I could take the old version of Wikipedia:Brilliant prose that exists in the history of WP:FAC, delete the redirect and replace it with the old content. But this is unnecessary; the old content could be linked in a brief snippet; though perhaps an example would be illustrative. But such an example would be misleading, because it would suggest. I will say that brilliant prose is not the successor to the featured article process; it's more accurate to call "featured articles" a rename of "brilliant prose". They are not really literally two separate things, but only philosophically two separate things. Brilliant Prose to Featured Article is not the clean break where FA became peer reviewed; it was renamed for aesthetic reasons. Therefore, the old name doesn't mean as much as I thought it did. But the old name could still contain discussion about the historical FA process and how it changed to peer-review, since it clearly changed since its brilliant prose days. If restoring brilliant prose, I should ensure all previous redirect links that would have linked to brilliant prose work. Concerning Wikipedia:Historical archive/Refreshing brilliant prose (which was basically FAR, something that is not clear from the title): "Archiving" this page is weird (expecially because the This leads me to another issue, I also think Wikipedia:Historical archive should be investigated. Ironically enough, the historical archive is a historical way of accomplishing archiving; I think archiving of old pages is currently mostly done by template notices now. It should be deprecated formally, and since I plan to do a mass move outside of the subpages for WikiArchaelogy (if they haven't been usurped), an MfD would be ideal to smooth that along. I don't plan to remove the redirects though, because there's no compelling reason to delete them; they should remain as a reminder of the historical archive's existence. I discovered BP through {{Article history}}; there's a link to Brilliant Prose on {{Article history}}, which is not the ideal target; it should be Wikipedia:Brilliant prose. There also doesn't even need to be a separate |
Cross-namespace redirects tangent
|
|---|
|
This is a red herring that I confused myself by including: I think we should discourage cross-namespace redirects, even for historical reasons; if one is interested in history, they can observe the link to the RfD in the deletion log entry (if the page is undeleted for content reasons, then a hatnote can be added). Historical preservation is not sufficient reason to tolerate namespace pollution. But this issue goes beyond a single page; there are currently around ~1200 cross-namespace redirects.
|
Merging
[edit]I would like to eliminate help and information page redundancy, but it's a bit difficult, because one should never lose content in doing so; one should simply compress and merge it. It is always useful to check page views to determine how accessible each page is before going ahead with a proposed merge.
- Help:Categories should be merged into Help:Category and redirected there (Help:Category has much more complete information). I suppose "Categories" might have some info that "Category" doesn't, and that can be evaluated.
- Wikipedia:Uploading images and Help:Uploading images (which is a redirect?)
- Wikipedia:Pageview statistics: Why is the link in the header template a button? It's unnecessary and misleading. Wikipedia:Web statistics tool could be merged into this. It was originally created to ... but is now unnecessary for ...
- Perhaps Help:Magic words should redirect to mw:Help:Magic words. But actually, many magic words allowed by Mediawiki have been replaced by Lua templates on Wikipedia. The Mediawiki page is already promintently linked in the lead of the Wikipedia help page.
AWB help page
[edit]The regex page should also mention more explicitly that it uses .Net regex, rather than relegating that information to the infobox. It could also stand to mention where regular expressions are used in AWB.
Add possessive groups to the Wikipedia AWB regex page.
From my tests, atomic groups `(?>` are supported, but not possessive quantifiers `+`
Possessive groups are a useful tool for limiting the repetitions of a [what's it called, wildcard, bleh] quantifier.
For example, the search `\[\[[^\]|]+\|` searches for a wiki link with a pipe. But this search could be done more efficiently as `\[\[[^\]|]++\|` to avoid the `+` quantifier reducing (since x
(Pretty sure this advice is on rexegg in particular: emphasize distinction.)
(Could add a picture of one of those black and white classic images with the caption, Regexes can often benefit from exclusive boundary conditions" or some text like that (but it might disturb the formality of the page, so perhaps not)
(Could also add this as a potential improvement to the spell check page.)
https://www.regular-expressions.info/possessive.html
Project pages miscellanea
[edit]- Help:Category should explicitly mention that it uses Unicode collation to sort redirects.
- "It affects only " -> "It only affects"
- This page is not detailed at all about how it's actually sorted... presumably ASCII with the addition of Unicode collation, favoring small words before large. This needs to specify the sorting mechanism more clearly, perhaps with examples.
- Rdcheck: Perhaps this should be mentioned in the parent help page itself, if it's not mentioned already.
- Wikipedia:Maintenance: "Things to watch out for" has a negative connotation, while some of the items in the list like "Helping editors" is not something to "watch out for".
- It'd be nice if
{{namespaces}}was included on all of the pages it linked to, but it does sometimes interfere with other sidebar templates like{{Guideline list}}.
Phabricator tickets
[edit]Perhaps I should have a section listing Phabricator tickets I'm interested in, ranked by how interested I am. But I don't know how productive listing by interest is, and I think I'll know the ones I'm interested in without looking because they're the ones that impact me the most.
AWB annoyances
[edit]- T99326: Undo and Ctrl+Z in the Replace Special window operate on the rules list exclusively, never the "Find" and "If" tabs
- T150795: Database Scanner - show the page source
- T222784: Show search snippets in Database Scanner search
- T222791: Add right-click menu option to open redirect directly in browser
Miscalleneous existing Phabricator tickets
[edit]- T5324: On redirect pages, "article" tab in top bar should lead to nonredirected page (&redirect=no)
T223701: Recent search regression: changing certain components in the search bar changes the URL to a plain index.php- T223709: Namespace selection in AdvancedSearch is not respected any more
- T223990: Automatically moving user talk pages during renames should not trigger "You have new messages"
Searching
[edit]I don't want to waste time reporting these on Phabricator if there are other search tools that satisfy these functionalities.
- Regex doesn't work in titles, e.g.
intitle:/^U\+/; but there's a separate (very slow) tool, Article title grep, for that.- I've developed a Quarry query that's hopefully a bit faster: query 36144
- I didn't realize the main search query, which could include quotations, or
insource:specifiers had to be beforeprefix:; this may be a search bug. Here's my examples:- "E^pi*i batch" prefix:"Wikipedia:AutoWikiBrowser/CheckPage" works, but prefix:"Wikipedia:AutoWikiBrowser/CheckPage" "E^pi*i batch" doesn't.
- insource:/E\^pi\*i batch/ prefix:"Wikipedia:AutoWikiBrowser/CheckPage" works, but prefix:"Wikipedia:AutoWikiBrowser/CheckPage" insource:/E\^pi\*i batch/ doesn't.
- Mediawiki search doesn't support atomic groups. It would be nice if it supported atomic groups. But this is probably more a PHP regex problem than a Mediawiki search problem.
- If one wants to search for two separate words in a title, they will have to enter 'intitle' twice in their query. But this might not actually work. From more recent experimentation, I believe it does work, though.
- Regex searching should have a way to set a limits on the search results processed (or rather, search in general should have the ability to set search limits, for the benefit of regex searching). But I guess even "processing" 20 results is more expensive than applying a regex to 20 pages. Still, it seems a waste to query an excessive number of pages if one can only look at 20 results at a time (or up to 500 at a time with the relevant settings.)
- Results 1 – 20 of 228,209
- Results 1 – 20 of 96
- I wish links to redirects in searching would at least include a direct link to the redirect (I can probably program a user script to allow this.)
Diff oddities
[edit]I've observed plenty of strange diffs. The below could be converted to a table. I can display the diffs visually on a page to show the differences. There must be a template for this. Ideally, it would use the same code (or at the very least, allow the same HTML) as actual Mediawiki diffs. I could make a user script to convert a diff to a templated version. It could be static, because on-demand is unnecessary (diffs don't change after they occur), and more intensive and coding-heavy than necessary. I can make up my own set of diffs on my test pages based on the weird diff handling I've found; this would lessen the risk of incidental revdel. I should also screenshot diffs.
- Special:Diff/15847306/16539802
- Why is there a spacing subtraction addition shown in this diff? I know it's clever that it can detect the moved text (it's probably a newline detector), but it would be nice if it disregarded the nonexistent newline change.
- Weird bracket diff handling:
- Special:Diff/833722887/835538082
- This is strange diff handling, to my eyes.
- Table editing display should be handled better than this. It should group the entirety of the new row together, instead of wrapping in the last cell of the previous row.
To report
[edit]Rights changing
[edit]This form of reporting is terrible because it lists all the rights before and after. It should only list the changed rights. A Phabricator ticket if I ever saw one (it probably already exists) It isn't normally necessary to remove multiple rights at once. Or to add and remove rights at the same time. I understand the edit summary is supposed to summarize rights changes, but that doesn't mean the log itself has to be overly verbose. There's one counterargument: one might want to determine one's exact rights at a certain point, but I think this is specialized use case, and should still be relatively easy to reverse engineer at a certain point. And this could be listed in a show more details tab for the log message.
It could just be summarized by a remove and addition list. Like "[Bureaucrat] added [Editor] to group [group], removed [Editor] to group [group]."
I wonder if I can code my own user script for this; I would like to have a shot, so I can show an example and understand the inherent issues better.
Headbomb's rights log, an example of how verbose the log is
|
|---|
|
Unsorted Phabricator tickets
[edit]- I found it curious that the file link here led to an "upload this picture" forced page redirect. So I investigated further and found that 'Image:' links behave differently than other links. I'm sure this is known, but it seems like it should be changed (I'm not completely aware of 'Image''s history as a namespace, and I'd assumed before now that it was an alias for 'W:', but this here proves that is not the case.) This could probably be turned into a Phabicator ticket (this has sufficient justification; basically, images should link to the actual image, not the upload page so one can know why the image was deleted.)
- I remember where I saw such a link; it was on an old featured images proposal.
- In the latest version of Firefox, anchor links put me slightly above the intended target. But this was only on one page, so I'm not sure what the deal; this doesn't happen elsewhere. But once again, I lack a specific example.
- Couldn't the interface read "Minor edit" instead of "This is a minor edit"?
- The notifications tab takes too long to pop up. Could there be an option where it would only load the most recent. I know it's not a network speed issue for the single most recent notification, since it displays the "1" notification bubble soon after the notification is sent.
- Interesting, I guess Special:RelatedChanges doesn't include changes done to previous titles, even though every other part of the Mediawiki software does? Example: Special:RelatedChanges/User:E to the Pi times i doesn't show my changes to User:E to the Pi times i/Tooltips before it was renamed to that title.
- Can we please have temporary passwords that are more secure than 12 characters? I'd feel more comfortable with 20.
- I wish Wikipedia used more AJAX. Specifically for going between consecutive diff pages, it should be much faster.
- Although I probably underestimate the difficulty of transitioning in that way, since I'm not really a Javascript coder. And I think the German Wikipedia already introduced such a tool. But really, I've seen quick AJAX, and Wikipedia is certainly not it. Perhaps that would require caching and pre-loading the next diff while one is still looking at the current diff. I should make a Phabricator ticket for this.
Miscellanea
[edit]Recognized good books from previous discontinued efforts
- This could be better summarized in a table and a brief explanation of "Book of the Month" and why it was discontinued.
- I need to find the page this refers to.
- User:CorbieVreccan/Admin Toolbox should probably not have Category:Wikipedia edit warring
Please report any bugs on the talk page of the script's creator.
Why not link directly to this page? The script creator will never change; the script maintainer might. Maybe "maintainer" was meant instead of "creator".- Why doesn't the first bullet point work in User:EEng#Museum of Wise Words? I know there are templates to fix it, but I'm curious why it doesn't work in the first place.
- WP-VG Nintendo task force / https://xtools.wmflabs.org/articleinfo/wiki.riteme.site/Donkey_Kong_Country#assessments
- Weird, especially since 'Nintendo task force' has never been linked as 'WP:VG/Nintendo task force' (it's 'WP:VG/Nintendo') (in both cases, with 'WP:VG' becoming 'Wikipedia:Wikiproject Video games'). I'll probably have to dig into the source code for wmflabs to find the answer.
- It'd be nice if I could get my IP cloaking request done. I will probably switch to the name "RetroWikiEditor"
- User talk:John Reid: Very interesting notes about tagging proposals.
- User:Quentin Smith/Admin coaching/3: Something interesting... All 3 letter abbreviations. Although it did stop. I don't find this organizational method particularly meaningful; I would look at some combination of page visits and ingoing links to determine important categories to study if one wanted to improve their editing of Wikipedia.
- Special:AbuseFilter/894
- 'AFCH' in the edit summary can allow people to ignore filters (but in the example case of self-published sources, it doesn't really matter because it's discretionary).
- I don't understand why this doesn't show up in the page's history. What's going on? I thought the filter only tagged the edit, not giving a warning or anything.
- Special:AbuseFilter/932 - could probably be merged into an existing filter.
- I wish there was an iterative tech news, so there's less redundancy between consecutive newsletters.
- Wikipedia:Redirects for discussion/Log/2017 January 2#C‑SPAN: I disagree slightly with the keep; that same argument would apply to almost any use of a hyphen, but this is not worth fighting over.
- Category:Television series: should probably never contain pages, not just rarely. But check relevant policies first.
- I wonder if a bot can consistency check this somehow. Unfortunately it seems difficult to determine and unlikely to be implemented in automated way anytime soon.
- File:Pasdecabale.jpg: The text recognition and tagging here is an interesting feature. I wonder how this works, and what other files it's on. It seems like it should be used on many files.
Questions
[edit]- I wonder how many pages are not indexed by robots and why. Presumably most articles from the mainspace are indexed; I think AfD is not indexed, and I wonder if any articles aren't indexed. How about user pages?
- Why doesn't deletion show up in some pages that were deleted (or is it all); unfortunately I saw this on one page, but I currently can't find it.
- I'd be curious to see a study of how often users cite essays, compared to how often they cite policies and guidelines. Then there are also explanatory supplements.
GA/FA bullet icons
[edit]This is among the most frivolous of my thoughts here, but I am looking for a format of GA/FA bullet icons for when I promote articles to GA/FA status. I realize it is premature to think about this, and do not plan to actively search for them; it's just a general note to self and a centralized storage place.
| Page | Description |
|---|---|
| User:Cirt/Quality improvement | This uses straight images and links using {{click}}, rather than using a situationally more suited template. I would prefer a more dedicated template. |
Template help pages
[edit]- This was previously a subpage named "Help/Template usage" (the parent page "Help" was devoid of content)
Parameter usage:
{{{Parameter name}}}{{{Parameter name|Default value}}}
Ordered parameters are named with consecutive numbers starting from 1.
Alternate names for the same parameter are often nested:
{{{Parameter name|{{{Alternate parameter name|...}}}}}}
AllPages nonsense
[edit]Why are both AllPages and AllPages with prefix necessary? All pages with prefix is clearly a more powerful utility, and its functionality could theoretically be merged into allpages
The only difference is the number of columns in the display. AllPages displays with 3 (4? 5?), while All pages with prefix uses an ordered list.
https://wiki.riteme.site/wiki/Special:PrefixIndex?prefix=&namespace=4&hideredirects=1
It would be nice if it had a hide disambiguation function too, but some things are too much to hope for.
Regex
[edit]I enjoy using regex too much, even when it's not necessarily the best solution for a problem. In the future, I'll probably transition to using C# in AWB. Here's an example of my silliness; this is a nonsense way to do this task, but I've created a regex for Wikipedia export metadata: \n <page>\n <title>([^\n<]+)</title>\n <ns>[^\n<]+</ns>\n <id>[^\n<]+</id>\n <revision>\n <id>[^\n<]+</id>\n <parentid>[^\n<]+</parentid>\n <timestamp>[^\n<]+</timestamp>\n <contributor>\n (?:<username>[^\n<]+</username>\n <id>[^\n<]+</id>
Tooltip exclusive regex searcher
[edit]\{\{\s*(?:[Tt]emplate:)?[Tt]ooltip\s*\|[^{}]{50,}\{\{\s*(?:[Tt]emplate:)?[Tt]ooltip\s*\|[^{}]{50,}
Embedded wikilink handling
[edit]Could be implemented with recursion, but meh. I do want to get myself refamiliarized with recursive regex, so maybe I will.
\[{2}File:(?:[^[\]]*(?:\[{2}[^\]]*\]{2})?)*\]{2}
Signatures
[edit]Notes
[edit]- ^ I am currently indefinitely topic banned from
policy and guideline pages and their talk pages as well as simply discussing changes to them anywhere on the English Wikipedia
. Commenting on policies and guidelines here could possibly be used as a justification for blocking, so I have avoided commenting here. If I appealed the topic ban, this would no longer be problematic. - ^ Technically, there was a tweak where I replaced
|with|at some point after it messed up some TfD formatting, but that didn't change the normal display of my signature. - ^ I originally mislinked my talk page as
User talkRetro.