User talk:AperfectBot

From Geohashing
Revision as of 21:03, 14 January 2010 by imported>Aperfectring (A format for the bot)

Archived discussions are here.

Purpose for this page

I feel it important to state what the purpose of this page is. It is intended to be a free and open forum for ideas about, criticisms on, and talking of the bot for automating Geo Hashing:Current events. Feel free to make any comments here, or if you prefer something less public, PM aperfectring on the IRC channel. --aperfectring 19:58, 18 June 2009 (UTC)

Bug Reports

Please notify me of bugs in the bot's output here. Also, please keep the <noinclude> tag below any active bugs, and above the "Fixed Bugs" header.

Minor parsing issue

The bot doesn't understand that the "and" on 2008-08-03_52_6 is between participants. - Jim

We will add a test case, and fix it relatively soon (we have a backlog of test failures, but this might get moved up due to its simplicity). --aperfectring 15:58, 10 November 2009 (UTC)

This page Template:Expedition_summaries/2009-01-31 is showing the AperfectBot/UpdatePage(sp?) text Jiml 20:57, 31 December 2009 (UTC)

The 2010-01-01 Page says that the Montreal hash was done by "someone", but it looks like a good name on the expedition when I look at it. Jiml 00:02, 5 January 2010 (UTC)

Malformed output

This page 2008-06-14_32_-117, is producing funky output on this page 2008-06-14. I think it is having parsing trauma with the page. I can fix the page, but I figured I'd let you take a swing at the bot first. Jiml 17:56, 15 October 2009 (UTC)

I will look into it, but my general thought on those older pages is to just leave them alone. I don't think there is a whole lot I can do about the malformed output there without breaking desired functionality elsewhere. --aperfectring 21:36, 15 October 2009 (UTC)

2009-11-24 Huh?? Portal? I think I confused it somehow with that page, but I'm not sure how.... Jiml 15:56, 25 November 2009 (UTC)

I think the ape has been getting confused a lot recently. Haven't gotten the chance to fix it quite yet. --APR

I think I've got another data point for you. It seems to get confused with a completely blank expedition template. Maybe it is looking for data in a section, but the section is empty. ("And good old bot's were reading wiki's and would cry..." Never mind) Jiml 19:22, 29 November 2009 (UTC)

Could be, but I thought I had it rigged up to ignore empty sections. ("Singing this'll be the day that I die u'Unparseable section';" Sorry, just had to continue the song) --aperfectring 22:27, 30 November 2009 (UTC)

Not updating in October

APR - the Current events page isn't getting updated. It still ends with Oct 6 for now. Jiml 05:30, 10 October 2009 (UTC)

This has been confirmed. I will fix it when I get home from work this evening. Likely its just that the bot isn't running. --aperfectring 15:42, 12 October 2009 (UTC)

APR restarted the bot, and all looks better. Jiml 17:56, 15 October 2009 (UTC)

This was due to the new way I am running the bot in a cron job. There is still a code fix needed here to prevent this from happening again. Basically, when I lost power (I am told this rarely happens in the Portland area, but it happened within a month of me arriving), the bot didn't clean up its lockfile. When I turn the computer back on, it was continually seeing the old lockfile, and deciding that it shouldn't start the bot. --aperfectring 21:36, 15 October 2009 (UTC)

Unreproduced Bugs

This section is for bugs which were not explicitly fixed. The symptoms of the bug don't seem to show up anymore, so there are two possibilities: 1) The bug was fixed with other updates to the bot, or 2) It was a transient error which repaired itself on the next pass of the bot.

Line Concatenation

Currently on Current Events it has:

2009-07-05

"McMinnville, Oregon - Jim, 3riador - A patch of woods close to Pumpkin Ridge Golf Course, in North Plains, Orego...

Davenport, Washington - Someone is, why not join them? - A farmer's field, 9.5 km east of Davenport.Courtenay, British Columbia - Elbie, Rhonda, Xore - Side of Inland Island Hwy, between Courtenay and Cumberland. "

For some reason the Courtenay line is gettting concatenated to the Davenport one. Jiml 22:24, 5 July 2009 (UTC)

It looks like that was a transient error. I don't know why it concatenated those two together, but I will look into it when I get home. --aperfectring 11:54, 6 July 2009 (UTC)

Fixed Bugs

External links

Current events has for 2009-07-11:

  • Palm Beach, Florida - Dtobias - In a driveway between the street and the sidewalk, in a residential neighbo...
  • Newark, New Jersey - Zubenelgenubi - Propelled by a strong west wind, I was early for once and killed some time ...
  • Boston, Massachusetts - Sparkyb - Soule Recreation Center and Playground Playground. Awesome! http://maps....

The "maps" link has a little link pointer, but the link isn't set up right. I know the bot doesn't like links, but I think it didn't strip it correctly.

Now that I read the fixed bugs section, maybe this is just a case of the more general problems with parsing external links. Jiml 22:49, 15 July 2009 (UTC) (edited on 17Jul)
The bot should be adding links now, so its a bug in the inclusion of links --aperfectring 23:02, 15 July 2009 (UTC)
It looks like it is a bare link, meaning that the bot didn't recognize it as a link. --aperfectring 23:07, 15 July 2009 (UTC)
I am in the process of fixing bare links. It should be done when I get home from work today. --aperfectring 12:17, 13 August 2009 (UTC)
The fix seems to be in place. If there are any other instances of links getting mangled, please let me know. --aperfectring 00:06, 14 August 2009 (UTC)

Duplicate User Names

Also, user names are listed twice. --Meghan 20:55, 2 July 2009 (UTC)

I was aware of the double names, but haven't worked on the bot for the last week, so hadn't quite gotten to fixing it yet.
This should now be fixed, let me know if it is still a problem. --aperfectring 22:43, 2 July 2009 (UTC)

Badly Formatted User List

I think it is mostly a data-in situation but the bot is producing this report for 2009-07-02 "Roswell, Georgia - NWoodruff, This, I, As, She, I, Seconds, I, He, He, 1), 2), 3), I, He, He, He, There, I - On top of a building in Lawrenceville. I was unable to make it to the top ... " Jiml 15:01, 3 July 2009 (UTC)

Yeah, I've noticed the problem. Technically it is a data-in problem, but my opinion is that it is a bot parsing error that should be fixed. I don't want to enforce too much of a uniform style to the expedition reports, because I think they should be an extension of the geohasher(s) that went on the expedition. I will think on it over this weekend, and hopefully come up with a way to improve the user list creation. --aperfectring 17:37, 3 July 2009 (UTC)
The issue noticed with the example provided was that the user list was immediately followed by the expedition description. I noticed that the user list was all bullets, and none of the description was. I changed the user list generation to look at the first name, and see if it has a bullet or a tab, and if it does, only match future lines if they also have one. This gets rid of the problem with the example provided. Let me know if you see any more cases of this. --aperfectring 13:59, 4 July 2009 (UTC)

Exclusion of Links

Looks like the bot doesn't like external links, as seen in the summary of 2009-07-01_49_8. --Meghan 20:55, 2 July 2009 (UTC)

The bot (currently) purposely deletes all links, because they are tricky to detect the actual length of. In keeping the length of location summaries constant, it was cutting the ends of them off and producing illegible summaries. I'll put it on my todo list to be able to interpret links correctly in summaries.
Most of all, thanks for the input! --aperfectring 22:04, 2 July 2009 (UTC)
Alright, I think links should be working in the descriptions now. Let me know if you see anything else wrong. --aperfectring 20:45, 4 July 2009 (UTC)

Images Being Included in Location Description

It looks like pictures are being included now, since I added links in. I will have to make a special case for removing Images, and probably categories as well. --aperfectring 11:54, 6 July 2009 (UTC)

Fixed --aperfectring 23:04, 6 July 2009 (UTC)

Not updating in August

Aug 10 note from Jim: The current events page still doesn't say anything about the McMinnville expedition/planning. (Jiml 18:15, 11 August 2009 (UTC))

2009-08-11:

--Ekorren 17:12, 11 August 2009 (UTC)

Hrm, that all is quite curious. My best guess is that it has been timing out periodically. The availability issues we all notice cause the bot problems as well. Template:Recent expeditions is pretty much the last page to be updated, so if it crashes out before that, it will not update that page. For the individual day templates, it should update them in reverse chronological order (e.g. 2009-08-11 then 2009-08-10 and so on). I know that the bot has had issues with writing to date pages in the past, so perhaps it started to have those problems again. I will try to look into what the issue is when I get home. I may not actually find anything, though, as it looks like the bot has updated the pages since you mentioned the problem. I apologize for the delay in response, but I have been busy at work, and just got back into town. --aperfectring 21:01, 11 August 2009 (UTC)

Based on what I saw last night, it seems like it is the inability to write to date pages (e.g. 2009-08-12). I've run into the problem before, but unfortunately, my solution was to just edit the date pages by hand. The last time this problem cropped up was the last time the slowness on the wiki became very noticeable, like it is right now. --aperfectring 11:56, 12 August 2009 (UTC)

In an effort to help fix the issue where older dates and the recent expeditions template are not updated when failures occur, I will take the following actions:

  • Adjust the order or writes to put the pages which are most likely to change to the front. This means the recent expeditions template and expeditions summaries templates for "today" and "future". Then it will be older expeditions summaries templates, the date pages, and then manual update requests last.
  • The other action I would like to take is to look into handling the write failure, so it doesn't cause a bot crash.

--aperfectring 12:55, 13 August 2009 (UTC)

Instead of the above fixes, I decided to only update date pages if the don't already exist. This should alleviate any issues with writing to date pages, except around 9:30 AM US Eastern. However, in my experience, very few people make changes to upcoming/past expeditions in that time, even though it is a time when many geohashers are aware of things. I am leaving this open for now, until I am sure that this prevents the majority of bad things. --aperfectring 00:09, 14 August 2009 (UTC)

Comments

Place general comments and praise for the bot here.

Suggestions

Place suggestions for improvements to the bot here.

Next/Previous

I would be interested in a feature that automatically links meetup pages together. e.g. at the bottom of the page, "previous" and "next" meetups in this graticule. For easy browsing without Back button. --Davidc 21:50, 11 August 2009 (UTC)

Revisit old pages

We might want APBot to revisit old pages from time to time, or if they "change", in order to catch updates to the pages that feed into the expedition summary pages. I re-ran it for Aug and it had outstanding changes for at least 5 old pages. Jiml 04:53, 26 August 2009 (UTC)

I was trying to keep the bot as simple as possible. As you now know, if you have looked at the source I sent, it isn't really all that simple even now. I could have it check for new changes since the last run, and update any pages that fit that, but that is complex, and error prone. I can look into doing it, but it will take some time, and I won't get started on it for a while.
As for scheduling updates to past dates, there are a lot of old dates, and they don't change very often. That's why I took the route of having users request updates if they add/change an old expedition, or want to make sure an old date is current. Once again, it is very possible to do so, but would slow down the already not very fast updating of newer pages, so I am inclined to leave it as is.
I am not against changing to either of those options, if people think it is a change for the better. --aperfectring 12:38, 28 August 2009 (UTC)

Inclusion in the Main Page gallery

Based on the reactions of people, I have reverted the Main Page back to its original state.

After inclusion

The summaries have been added to the front page, since people expressed a desire to see what it looked like before deciding either for or against it. Use the sections below to express your opinions now that the summaries appear on the front page. They have also been updated to have a "More summaries..." link when more than 4 would appear on the front page.

Opinions

For voicing your Support (keep the summaries), Oppose (remove the summaries), Do Not Oppose (no opinion), or Needs Work. Needs work implies that the summaries need to be improved upon before you feel they are ready for the Main Page.

DNO - I still kinda lean against including them, as there is some duplication, and it makes the gallery kinda long, but I also don't care enough to make it a firm opinion. --aperfectring 13:41, 13 July 2009 (UTC)

After more soaking, I am changing to oppose. I really prefer it as a gallery there. --aperfectring 16:52, 16 July 2009 (UTC)

DNO - Limiting the number shown is a huge improvement. -Robyn 21:54, 13 July 2009 (UTC)

support - ack. -- relet 12:09, 16 July 2009 (UTC)
support - I like them, and think the planned exped. summaries in particular are handy to have a link to on main page.--Eserchie 12:16, 16 July 2009 (UTC)

Oppose - I think it just displays the same information as the pictures, except in a less interesting way. The gallery is a cool way to quickly see what everyones been up to, and the text takes away from that by making it longer, less interesting and repetitive. I like the expedition summaries on the current event page, and I'll go there when I want to see them. Thepiguy 13:44, 15 July 2009 (UTC)

DNO - leaning towards oppose, but I never visit the main page. Maybe leave a link to the expeditions section on the date page? Like "List of expeditions" --joannac 12:17, 16 July 2009 (UTC)

  • DNO - I prefer to have only pictures on the main page, because for me it looks cluttered with the text summaries, but I don't want to block either. - Danatar 13:50, 13 July 2009 (UTC)
  • Oppose. I needed to be pointed to the main page to actually see the changes (as I usually don't visit the main page at all). I think it takes up too much space without really being that interesting, and it generally looked much nicer when it was a compact gallery block as opposed to several distant rows of pictures. This could be ok for a monthly or weekly overview page, but not for the main page, where it's not about giving a complete overview but about "hey, look, that's what we do". --Ekorren 16:54, 16 July 2009 (UTC)
It seems that we also need a discussion on what the main page should be about. To me, it has always been the first page to check if there has anything happened on the wiki. I would often not notice a report unless it has a picture on Main Page. It should introduce people to the sport, but it should also be the a place to check what has happened when you haven't been able to read the recent changes for a while. -- relet 17:21, 16 July 2009 (UTC)
This purpose is better served by the existing monthly pages anyway. --Ekorren 17:28, 16 July 2009 (UTC)
Only you have to pass by the Main Page to reach them in the first place. ;) -- relet 17:40, 16 July 2009 (UTC)
Really? So I seem to do something totally wrong by just changing the filename in the URL from whatever page I am at the moment... :P --Ekorren 17:48, 16 July 2009 (UTC)
This kind of talk is probably best suited on the Main Page talk, but no one goes there anyways, so here is fine. Very few people visit the Main Page, it seems, so I tend to see it as more of an introduction to the activity, rather than a personal jumping off point. I think the current events page is turning into more of a "what's been happening recently" place to go. --aperfectring 17:53, 16 July 2009 (UTC)

Oppose. The galleries looked far better, and we do need far less text on the main page. I liked judging books by their cover (despite the fact that I read most reports anyway :)). Ephphatha 18:11, 17 July 2009 (UTC)

Needs Work. "..." halfway in a sentence, or even after two words is crazy. If expedition writeups have a short summary in them that's tagged like that, maybe that will work, but this isn't. Something like a max of 140 characters, that's tagged [summary], and if it isn't there, then no summary will be on the front page.--Arvid 18:17, 17 July 2009 (UTC)

Comments

For discussion of the issue.

Robyn, on the maximum number to include. I take it that the maximum number should only apply to front page inclusion. Is this correct? I should, through some template magic applied at bot-time, be able to swing that. --aperfectring 19:16, 13 July 2009 (UTC)

thepiguy's point is very well taken. Today however there are quite a few expeditions that are in the written summaries but don't have pictures. The written summaries show us expeditions by users who

  • are too shy to post a picture
  • take a week to upload pictures
  • haven't returned yet, but who made a page before they left
  • don't have a camera

I wouldn't be disappointed if the written summaries were to disappear from the main page because it IS messier and we CAN go to Current events for them. -Robyn 07:12, 16 July 2009 (UTC)

I'm still not a fan of the text summaries. About duplicates: we could leave out the text summaries for expeditions which are included in the picture gallery, so that only the no-pictures-(yet)-expeditions are describen in text. - Danatar 12:33, 16 July 2009 (UTC)

Initial inclusion

The expedition summaries are rather stable, and provide a decent summary of what is going on for a particular day. Since that was part of the vision of the Main Page gallery, should these expedition summaries be included on it? Inclusion in the main page galleries shouldn't be very difficult.

Opinions

For voicing your Support, Oppose, Do Not Oppose, or Needs Work. Needs work implies that the summaries need to be improved upon before you feel they are ready for the Main Page.

  • DNO - I like the front page as just a gallery, but am not too strong on the opinion. --aperfectring 13:06, 9 July 2009 (UTC)
  • DNO slightly to the side of oppose, but don't want to block. -Robyn 15:47, 9 July 2009 (UTC)
  • support - I liked the summaries on main page back when they were still written by hand. Bot seems to be stable enough to need only minor fixes, which become much more apparent when they are on Main Page -- relet 16:24, 9 July 2009 (UTC)
  • DNO I find the main page a little cluttered as it is, but I'm not passionate enough about it to oppose. --Meghan 17:28, 9 July 2009 (UTC)
  • DNO Mainly becauth it's worth giving it a go to see how it works. we can always change our minds if it turns out to not work. --mykaDragonBlue [- i have no sig -] 06:13, 13 July 2009 (UTC)
  • DNO Let's see what it looks like. If it is too cluttered, we can stop it, and they will all go away in a couple of days. Jiml 06:27, 13 July 2009 (UTC)

Comments

For discussion of issue.

Planned expeditions

Right, so the advent of this has re-opened an old issue that just won't stay buried: short planning pages. There are many short-ish planning pages from when the algorithm started, as well as from new users. After discussion it was decided to move the short ones into the graticule page, and leave the ones with more than a sentence.

HOWEVER with the advent of aperfectbot, is it now again useful to leave these pages for inclusion on day pages, and to give a complete picture of activity on a certain day?

As an example, 2008-07-12 50 6 is a page I would normally leave as a line on a grat page, like the "Geohashes considered" section on Edinburgh, United Kingdom. But doing so means it won't turn up on 2008-07-12. --joannac 06:06, 13 July 2009 (UTC)

  • I assume that each successful expeditions should get their own page, so they can be counted in the various statistics.
  • And for planning purposes, the thought was that once it looks like someone was seriously thinking of going, an expedition should get a page so that it will show up on current events.
Do we delete/migrate that text if the expedition doesn't pan out?
  • We could have a "geohashes considered" section as standard on active graticules until some real planning starts to happen.

Jiml 06:21, 13 July 2009 (UTC)

  • I think that once someone is seriously going, they should create the page. After that, no matter whether the expedition pans out or not, the page should stay.
  • I also think that if you are seriously considering going, and take the effort to make a page and categorize it, that you should take the time to fill out at least a rudimentary plan. e.g. I am going to head down Fishermans Road as far as possible, then hike over land.
  • I have in my TODO list an item to look at specific graticule/graticule talk pages where planning regularly happens, and include any plans for the appropriate dates.

--aperfectring 11:52, 13 July 2009 (UTC)


Per-User Expedition List Planning

I am using this section for designing out the workings of creating per-user expedition lists.

Format Text Requirements

This is a list of all the possible data that should be able to be represented in the expedition list.

  • Date
  • Graticule (By address and name)
  • People met/encountered during the expedition
  • Description of the location
  • Mode of transport.
  • Coordinates reached/coordinates not reached and reason
  • Expedition #/Success #

A format for the bot

I think I'd like something along these lines:

  • reached_icon link - gratname - people (max 2, et al.) - transport_icon - usertext

Basically what I have on my page now. But I'm not sure I understood the format well. -- relet 20:52, 14 January 2010 (UTC)

I can help with getting it in the correct format. There will need to be some changes to the bot to get it to work in the manner you described, but for the most part I think they would be good changes.
  1. REACHED_ICON LINK - GRATNAME - PEOPLE:2 - TRANSPORT_ICON - USERTEXT
New tokens:
REACHED_ICON - should automagically put either the green or red arrow in place
PEOPLE:<number> - should print the other people who attended up to a maximum of number. They should be sorted in some manner, likely alphabetical, so that it remains stable.
TRANSPORT_ICON - should automagically put icons for the transport forms it detects.

--aperfectring 21:03, 14 January 2010 (UTC)