User talk:AperfectBot

From Geohashing
Revision as of 13:58, 22 June 2009 by imported>Aperfectring (Bug Reports: new section)

How about "Unknown, but someone seems to be planning it"? -Robyn 02:32, 16 June 2009 (UTC)

For which one? --aperfectring 02:37, 16 June 2009 (UTC)
Which part are you suggesting that for, the who or the where? --aperfectring 13:49, 16 June 2009 (UTC)
Sorry, that wasn't clear. I just want to be sure that an entry with unknowns looks more like "someone is going here, but the bot can't figure it out" as opposed to "you'll be all by yourself if you go here". "Unknown" is fine for the who, or "Unknown people." I suggest something like "no location description," "description unavailable" or even "please add a location description," for the where. I see that "why not go on a spontaneous adventure" is way friendlier. Maybe something of that sort, ignoring the lack of description and substituting something generic, like "Are you coming too?" would be best. -Robyn 20:57, 17 June 2009 (UTC)

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)


Conversation moved from Geo Hashing talk:Current events

How should it work?

Cool! In manually populating the page, I found it most productive to search for e.g. "2009-06-13". There were only two pages in the planning category, but there were things on graticule pages and graticule talk pages. A bot could probably link to the page and section it found that text, and paste the first line following it. Graticules like New Jersey that post the daily location even when no one goes could be either put on an exclusion list or prevailed on to include a string like "no plans yet" that tell the bot to skip this one. At any rate, when it is botted, there must be room to easily edit manually. How often would the bot look? -Robyn 18:29, 12 June 2009 (UTC)

I think the key is to influence people to create their expedition page before leaving, and make sure that the expedition page is added to the category. If we can get that mentality in place, then the bot can check for changes in pages that link to the category and only update the page as needed. Do keep in mind that this is all just vague thoughts right now, and I will plan it out much more fully when I am not at work. =P --aperfectring 18:39, 12 June 2009 (UTC)
This is so wiki. I love how I can have ideas and then people make them better and make them work. -Robyn 18:43, 12 June 2009 (UTC)
Oooh, I just had a flash of genius (it hurt). But I had better post it before I forget. We could have a "Have the bot create your expedition planning page for you" section where you give it your graticule's name or lat/lon, the date, and possibly a user (for who is going), and it will create the page using the approved template for you in a couple minutes. --aperfectring 20:28, 12 June 2009 (UTC)
That is in line with my My Vision of the Future, so I heartily approve this possible new development. -Robyn 23:02, 12 June 2009 (UTC)
BTW, if the page goes by planning pages, then it doesn't even have to crawl. It would be sufficient to link under each date to, e.g. Category:Meetup on 2009-06-14. But it would be better to have them listed right there on the page, especially if the bot would retrieve graticule names. -Robyn 00:57, 13 June 2009 (UTC)
My vision is to have, for each date, a list of expeditions in planning. Each of which would contain a link to the planning page, where the text is the graticule's name, the name(s) of the participant(s)t and a summary of the location. The last part will probably be the biggest challenge. People usually have a section where they list who is going (though not everyone signs up right away), but the summary of the locations that I have seen are usually a bit longer, and would need to be cut down. To get a graticule's name from its lat/lon requires looking at All_Graticules, but that shouldn't be very difficult. For All_Graticules, I think I will pull down a copy of the page daily, and run off of that, so I am not constantly trying to download that monster of a page. BTW, has anyone noticed that All_Graticules doesn't follow the naming conventions? --aperfectring 12:08, 13 June 2009 (UTC)
Moving All Graticules to All graticules was discussed, but rejected over concerns for the automation that uses it. I was thinking about this last night, that the "I'm going on an expedition" bot should solicit date and graticule (required) with the option for specifying participants and a one-line description. Perhaps there should also be a provision for specifying that it is a retro or alternate. As a first pass for listing already created planning pages, I would just grab a line of text. So what if you have a lot of "Hey guys this is totally doable" or "I wish I could go but I have to pick up my dog at th". At least you have links. People might even learn to start with a good tagline. -Robyn 16:47, 13 June 2009 (UTC)
I agree that the primary goal is to create the list. It should be trivial enough to include a user list. For the descriptions, if it is too long (some people are VERY verbose in describing the location, I have noticed), perhaps I should cut it off after 100 characters or so (something like 20 words), and if it is longer than that, add an ellipsis. The length isn't set yet, I will likely go with what looks best. I have used the same toolkit as User:ReletBot and gotten it to log in, and download the Main Page so far. I will continue working on it. For now I will hard code pages to look at/create a list for. User:AperfectBot has now been created. --aperfectring 17:23, 13 June 2009 (UTC)

Arbitrary Section

(because the sections were getting big to edit)

Using 2009-06-14_49_-122 as a test bed, and some code Relet graciously donated, I have come up with a bot which creates a line of text in the following format:

link - personA, personB, personC - Where it is and if it is long then I...

That looks awesome. Will the bot be confused if it gets an edit conflict, like I just did??
It doesn't do any editing yet. Most likely it will pull and put the page so quickly that no one will notice. If there is any error on write (which an edit conflict is), I will likely take a simple approach and just discard my changes, sure that I will catch it next time. On a side note, people should put their expedition planning pages in this format: Template:Expedition/Example --aperfectring 23:07, 13 June 2009 (UTC)

See Talk:2009-06-14 40 -104, as of its original creation time, as a perfect example of an expedition planning page that the ideal bot should be able to find. -Robyn 17:40, 14 June 2009 (UTC)

I will look into ways of parsing that, since I agree, that should be enough to create an entry. Currently the bot ignores talk pages, since the ones I saw in the category were the historical planning, and the expedition information was on the page itself. --aperfectring 15:08, 15 June 2009 (UTC)
When I get done with work today, I am going to work on being much more permissive about the location and user parsing on pages. --aperfectring 19:09, 15 June 2009 (UTC)
Vancouver sometimes plans the expedition on the main page and then moves the planning to the talk page when we're done, and sometimes plans straight on the talk page. -Robyn 03:02, 16 June 2009 (UTC)
Current cycle, which is finishing up as I type this, includes an update to finding users. If it can't find a user section, it finds all unique User:* tags, and uses that as the user list. The next cycle, which will begin right after this completes, and be done in 45 minutes or so, includes a fix to the user list generation, and includes an update to the location text. After those two, it should be fairly good at picking up stuff from the planning pages. --aperfectring 00:02, 16 June 2009 (UTC)
OK, so it took longer than I thought, and had a few bugs, but a new version is on its way, guestimate of 20 mins? --aperfectring 02:05, 16 June 2009 (UTC)
/me sighs at people intentionally doing stupid stuff. I think I have now got a good setup to work, we shall see in 30 mins. --aperfectring 02:44, 16 June 2009 (UTC)
People are doing stupid stuff to test the bot? Take it as flattery. -Robyn 03:02, 16 June 2009 (UTC)

After some talk with ekorren over the weekend, I feel that the bot should also do some searching of graticule pages (or talk pages) for planned expeditions. That part will require some work, and I will think about how to go about it. My current thought is that if your graticule tends to talk about upcoming hashpoints on your graticule page, then you can include your graticule in a list, which the bot will traverse, and parse your pages for info about upcoming hashpoints. --aperfectring 15:16, 15 June 2009 (UTC)

If you searched for all occurrences of "YYYY-MM-DD" for the given date, discarded the ones that linked to existing expedition pages and then just posted a link to that page with the line following, you would get most of it. Finding the graticule is harder if the link is on a user page, and you'd get messed up occasionally by things like Talk:Vancouver, British Columbia being the clearinghouse for expeditions in the whole area. Portland, Oregon does the same thing, I think. But there would be nothing that the user couldn't sort out by following the link. -Robyn 18:35, 15 June 2009 (UTC)
I have noticed that people will usually mention the different graticule in the location description, so it shouldn't be as big of a deal there. I think that plans made solely on a graticule talk page probably deserve to be in something akin to a "tentative" category. When people talk about hashpoints on the talk page, usually its to see if anyone else is planning on going. I would also advocate that if someone is making more definite plans (e.g. describing their bus route to a hashpoint), then it deserves to be put on its own expedition planning page, and linked from the graticule's talk page. --aperfectring 19:09, 15 June 2009 (UTC)
Agree, but what we should do and what we do don't always match. I like that this project is a combination between looking for planning pages and encouraging people to make planning pages. I favour erring on the side of having EVERYTHING on the Current Events page and letting the user who is considering it judge for themselves whether the link destination constitutes a definite intent or not. I can distinguish a lot better than a bot between "I wish I could do this one" and "leaving right now to do this one." -Robyn 02:37, 16 June 2009 (UTC)

Playing Nicely with Users

This looks very good. Harder for the user to screw up that I thought. Both the "add below this line" and the "This page is automatically generated. Any edits to this page will be overwritten by a bot." are very clear. Somewhere it can also tell users that iff they want to change what the bot displays, to change the first 75 characters of ... which section does it look in first? Does it display 15 names for the participants, or just the first n?

If I edit above the line I just get overwritten, right?

Actually, I wrote it so that it should play very nice with users. Even text above that line should be OK. Only the date headers will be forcibly replaced. There is only one thing which is taboo, and that is equal signs. I need to put a comment in about that. I will write something up about how to change what is displayed for the auto-generated text. Most likely I will update what I already have up so that it is current. --aperfectring 19:27, 21 June 2009 (UTC)

Testing

Next step: pull all pages from Category:Expedition_planning and parse them, producing a list, and thus verifying my algo, and its error handling. --aperfectring 22:43, 13 June 2009 (UTC)

Update: The output from the test runs will be put here: User:AperfectBot/Test_Page

Bug Reports

Please notify me of bugs in the bot's output here.