I'm working on modernizing Rosetta Code's infrastructure. Starting with communications. Please accept this time-limited open invite to RC's Slack.. --Michael Mol (talk) 20:59, 30 May 2020 (UTC)

User talk:ImplSearchBot

From Rosetta Code

Did you notice that "Unimplemented tasks by language" is on the list for all languages? --Mwn3d 19:49, 19 February 2009 (UTC)

Which page are you referring to? --Short Circuit 20:08, 19 February 2009 (UTC)
Every "Tasks not implemented in" page. "Unimplemented tasks by language" is part of the list. --Mwn3d 21:39, 19 February 2009 (UTC)
It's because Category:Unimplemented tasks by language was accidentally included in Category:Programming Tasks. I fixed it. Next time the bot runs, it'll clear up. --Short Circuit 22:01, 19 February 2009 (UTC)

The "==End of List==" part of the page needs a new line in front in case the list is empty. --Mwn3d 03:32, 20 February 2009 (UTC)

done. --Short Circuit 03:54, 20 February 2009 (UTC)

Omitted tasks show up in both lists. It seems like they should be in one or none doesn't it? (Ex: "Command Line Interpreter" in Reports:Tasks not implemented in Java) --Mwn3d 21:09, 23 February 2009 (UTC)

Indeed. That is a bug. I'll have it fixed tonight, but probably not before the 7PM EST update. --Short Circuit 21:27, 23 February 2009 (UTC)

The bot can't count[edit]

Right now, Reports:Tasks not implemented in Tcl says that there are -1 tasks not done in Tcl. While I admit being a little bit of a completionist, that's overzealous; I'm sure that there are tasks we've done but which aren't yet described, but I'm pretty sure that the bot shouldn't be counting them. —Donal Fellows 21:52, 5 August 2009 (UTC)

New page not marked as "new"[edit]

I noticed that ImplSearchBot created a page -- Reports:Tasks not implemented in PowerBASIC -- but it's not marked as new in the page's history. I don't know if that's a ImplSearchBot issue or a wiki issue, or if it even needs to be "fixed" at all. -- Eriksiers 18:35, 11 August 2009 (UTC)

No pages are marked as "new" in their own history in the comment when an edit comment is entered. No page is marked as "new" with an "N" in its own history. The pages are marked as new in the recent changes list:
(diff) (hist) . . Nmb Template:Unimp body PowerBASIC‎; 20:10 . . (+8,667) . . ImplSearchBot (Talk | contribs | block) (ImplSearchBot:Updating Template:unimp_body_PowerBASIC)
You'll have to "show bots" in the recent changes to see that. --Mwn3d 18:41, 11 August 2009 (UTC)
Ah, I see. Guess I'm not as familiar with wikis as I thought. -- Eriksiers 18:44, 11 August 2009 (UTC)

Improvements to output[edit]

  • The listings of tasks listed should use a template for each task, which provides a link to the task's edit page. (Or possibly even "edit new section".) --Kevin Reid 02:52, 14 August 2009 (UTC)
  • The bot could generate more useful edit summaries. Suggestion: link to pages if possible, e.g. "Removed Mutex, Arrays; added Seven-dice from Five-dice", and if they don't fit in the max summary size use the count of added/removed tasks instead. --Kevin Reid 03:29, 14 August 2009 (UTC)
Working on it. The solution is going to be a bit roundabout, but it'll happen. --Short Circuit 19:00, 23 August 2009 (UTC)
  • Suggestion: Under the Not Considered header, link to task pages with an edit link beside (e.g., in parentheses) as is done in the main list. Tayloj 18:40, 23 August 2009 (UTC)
I actually fixed that in the bot last night, but I missed a step in getting the changes over to the copy of the bot that runs and updates the wiki. I'm doing a manual run now; It should be fixed within about half an hour. --Short Circuit 18:58, 23 August 2009 (UTC)

Need special case for nested pages[edit]

I'd like to have a special case for large tasks like RCBF which keep the actual code on nested pages like RCBF/Common Lisp. Currently, this requires the main page to include the category for each language. In addition, each nested code page also manually includes the category for the particular language. This inflates the count for those languages which solve these nested tasks.

I'd prefer for ImplSearchBot to detect nested solution pages and only count them, so we could omit the language categories from pages which don't have code. (This would also catch a handful of other tasks which mix inline implementations and nested page implementations.) I'd also like for the root pages to just link to the appropriate category, like Category:RCBF, so we don't have to maintain both the nested solution code and a link to the solution code on the root page. --IanOsgood 19:56, 18 December 2009 (UTC)

ImplSearchBot has been dead for three months, now. Edit Template:Unimpl_Page to add/remove features. I will not be adding special cases to the MultiCategorySearch code, because that road leads to the maintenance headaches that led to ISB sucking up the vast majority of the time I had available to spend on RC this year. The only practical alternative I can think of is to change MCS's syntax to support arbitrary SELECT/subselect queries and subqueries, and that opens up a nasty potential DOS attack if it's not done very carefully. I have a large number of other backlogged things I need to focus on before i consider it. Poke Opticron if you'd like to take over further development of it; I think he's sick of it. I brought up various options at various times to deal with RCBF and friends several times, but since nobody commented further or took further action, we have what we have until someone provides something better. --Michael Mol 20:44, 18 December 2009 (UTC)