Wikipedia talk:AutoWikiBrowser/Dev/Archive 3

From Wikipedia, the free encyclopedia

Archive This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.
← Archive 2 |
Archive 3
| Archive 4 →


Contents

Sourceforge problem

Anybody knows what's wrong with Sourceforge? I often can't preview sources like here and the diff shows nothing like here. Jogers (talk) 14:57, 14 October 2007 (UTC)

The binary diff, ive seen it a few time before. But never been bothered by it. It just seems ot be random files now and again.
For the other, its viewVC or python errror. Its not just you Jogers :). /me thinks SF should update them Reedy Boy 15:09, 14 October 2007 (UTC)
Reported - [1] Reedy Boy 15:23, 14 October 2007 (UTC)
Its been replied to and closed.. See the link above. I suggest we normalise the text in the files... Reedy Boy 17:50, 25 October 2007 (UTC)
I've converted Tools.cs to UTF-8 and it can be correctly displayed now but we still can't preview previous revisions and the diffs. I wonder if there is any way to update them? Jogers (talk) 12:31, 26 October 2007 (UTC)
Get ViewVC to update the software to fix it.. And then get SF to install the update...? Reedy Boy 18:40, 27 October 2007 (UTC)
Never mind, I can always check the diffs in TortoiseSVN. Jogers (talk) 12:42, 31 October 2007 (UTC)

Feature Requests

What a backlog!!

Reedy Boy 14:01, 20 October 2007 (UTC)

Probably, some of them should be declined as too vague, needlessly specialized or not in line with AWB's ideology. MaxSem 09:54, 21 October 2007 (UTC)

Some dev should be constantly monitoring the Feature requests page as it looks really dead.--Kozuch (talk) 21:28, 14 April 2008 (UTC)

Legal fun

Eek! I've found a good C# diff library, http://razor.occams.info/code/diff/ . Then the copy(left|right|up|down|<censored>) fun begins:

Is there anyone alive? I'll just add it a a binary (which is obviously allowed) and start working on rendering. MaxSem(Han shot first!) 11:08, 19 November 2007 (UTC)
Yeah, i am... Just wanted to get time to go over the licensing and see if you had missed anything, or anything that may help us —Reedy Boy 12:33, 19 November 2007 (UTC)
Looking good Max. Also, seems quicker...? —Reedy Boy 19:13, 24 November 2007 (UTC)

Typo statistics

I have an idea, but it needs much polishing before we start implementing it. So: AWB should log all typos to some kind of log. Two things should be logged: typos found and false positives (when a regex matches, but proposed correction matches the word found). Several issues:

  • If we log it to a file, how should we prevent it from growing forever? Spamming users with requests to submit their data to us once their logs reach 50(?)kb? Automatically uploading them (bad idea)?
  • Where should they be submitted? (E.g. to subpage of RETF on every wiki?) What about smaller projects that use typos from en? (Are there such projects?)
  • How should be process the logs?
  • Is it really worth it?

MaxSem(Han shot first!) 21:14, 20 November 2007 (UTC)

Wikipedia_talk:AutoWikiBrowser/Feature_requests#Counting_Regextypo_fixes - Like that?
It'd be a bit difficult, i suppose to do it. If it was done to a subpage, AWB would have to parse it.. and we'll get edit conflicts and alike.
The simplest way, would be some form of submit to a Database somewhere, and have a webscript to make some statistics... But what would they submit? Original word, replaced word? As the regex's will change..
The "Is it really worth it", is the thing. Yes, its doable, and would be a major new feature. But what actual gain is going to come from it...? —Reedy Boy 21:32, 20 November 2007 (UTC)
It's needed for us to know which regexes are problematic and either should be fixed not to cause false positives or removed if it's too infrequent. MaxSem(Han shot first!) 21:48, 20 November 2007 (UTC)

Next Release...

4.2 i suppose... New stuff, changes, and the changed Diff library..

—Reedy Boy 15:46, 25 November 2007 (UTC)

Let's create a checklist. MaxSem(Han shot first!) 13:42, 4 December 2007 (UTC)
  • New functionality
    • Y Enhanced diffs
      • Seems pretty good. Need to try actually doing some normal AWB work and check it out better. —Reedy Boy 15:49, 4 December 2007 (UTC)
    • Y Some speed-related improvements, but nothing drastical
      • This i think, has just become a typical ongoing improvement for AWB —Reedy Boy 15:49, 4 December 2007 (UTC)
        • Startup has been seriously sped up, do I've marked it with a bug big check sign. MaxSem(Han shot first!) 14:42, 15 December 2007 (UTC)
          • Yup - Hugely faster! =) —Reedy Boy 21:56, 16 December 2007 (UTC)
    • N I've added some code to load site matrix, but it isn't used anywhere so far. Possible uses include site selection dialog and more sophisticated metadata sorter (no speed gain for WP but it's quite possible for other WMF projects with less languages available). MaxSem(Han shot first!)
      • Would be a good idea, then languages dont actually have to be added manually. I know we dont get many requests (well, not recently at least)... Maybe loading the Namespaces via the API/whatever for the wiki upon use would also be better than them all hard coded...? —Reedy Boy 15:49, 4 December 2007 (UTC)
  • Bug fixes
    • Y As always, lots of them
      • Yup, list is generally empty.. —Reedy Boy 15:49, 4 December 2007 (UTC)
    • N Emboldening code isn't fixed so far
    • Y We've gotta do something aboug {{reflist}} - related complaints
      • If there are cases it should stay (ie certain pages), and no actual consensus to replace it.. Shouldnt we really disable thiks feature? —Reedy Boy 15:49, 4 December 2007 (UTC)
      • rev 1875 - Disabled by MaxSem - We proceeding with it disabled then? —Reedy Boy 14:04, 7 December 2007 (UTC)
  • Y Testing - have anyone made a good run with latest version?
    • No, going to try now.. —Reedy Boy 15:49, 4 December 2007 (UTC)
      • I think now it's tested enough. MaxSem(Han shot first!) 15:39, 21 December 2007 (UTC)
        • Just tried it myself. Ran through my watchlist and did some typo fixing. Just that stupidly long inital processing time in my windows version that is a problem for me! —Reedy Boy 15:49, 21 December 2007 (UTC)
  • Y Just to add to the list - Vista/64bit stupid typo first processing time. Pretty sure i wont be the only person come release who has this problem...—Reedy Boy 16:59, 5 December 2007 (UTC)
    Vista sucks. MaxSem(Han shot first!) 20:45, 5 December 2007 (UTC)
    Haha —Reedy Boy 23:04, 5 December 2007 (UTC)
    Fixed as per bugs :) —Reedy Boy 22:03, 29 December 2007 (UTC)

Pre Christmas

Are we going to try and get the next version (4.2?) out before christmas? ie This weekend?

—Reedy Boy 16:04, 21 December 2007 (UTC)

Help for a script

moved to WT:AWB

AWBPackager

Now included in the Extras folder

Simple program, run it, follow the basic prompts, and it collects all the files from the various release build folders, and puts them into a temporary directory in the running directory, then zips it up based on the name given

Useful for releases, but also so we can quickly and easily build SVN versions and package them for users to try!


—Reedy Boy 12:12, 31 December 2007 (UTC)

AWB SVN Snapshots

Avaliable here

For users who are experienced with AWB, and want to use a more up to date version, but are unable to download from SVN themselves and build

May try and make the AWBPackager FTP upload to there...

Not very pretty, but it serves its purpose - Simple webpage to list all the FTP files in the folder :)

—Reedy Boy 12:15, 31 December 2007 (UTC)

GPL 3

Why is a copy of the GPL version 3 shipping with AWB? I don't recall any dicussions about this or giving my permission for my work to be relicenced (which I presume is needed although I stand corrected if not :) ) --kingboyk (talk) 20:25, 1 January 2008 (UTC)

Diffs

I'm finding the default diff in the current AWB (green) to be nowhere near as clear or obvious as Mediawiki (red). Would it be possible/desirable to make the AWB default diffs (CSS?) look exactly like Wikipedia's? --kingboyk (talk) 00:52, 2 January 2008 (UTC)

One other thing. I don't like the new buttons - why "Start processing" and "Stop all tasks"? "Stop" and "Start" are so much clearer... --kingboyk (talk) 01:03, 2 January 2008 (UTC)

Translations of AWB

Will you plan integration of translations into different languages ? For example, by moving character strings in parameters files. Thus, we could contribute to translation into different languages. You can contact me (email) for French translation. --Shawn86 (talk) 09:15, 3 January 2008 (UTC)

It could be done. Im not sure the best way to do it really, as there would have to be a lot of strings loaded and applied. Of course, as we can be used on multiple wiki's, we are open to suggestions as to how best to do this, and therefore implement it if we see it feasible! —Reedy Boy 09:36, 3 January 2008 (UTC)
Resource files isn't it? (Caution: I know nothing abt this topic). --kingboyk (talk) 14:54, 4 January 2008 (UTC)

Wiktionary & mandatory updates

Can the "mandatory updates" thing be restricted to use on en.wikipedia please? That was a fairly rude update for me.

For Wiktionary, I'm having trouble declaring myself as a bot. wikt:WT:AWB. Am I setting this up in the wrong place? The 4.1.4.2 AWB version throws a fit, maxing out the CPU when I try to log in now.

Can some of the warnings be customized? For Wiktionary, entries that do not start with a language heading should be marked as an error (for one small example.) This version, being heavily Wikipedia oriented, makes the opposite recommendation/warning.

Thanks in advance, --Connel MacKenzie - wikt 02:11, 14 January 2008 (UTC)

That's not a typo...

There are a few isolated cases where a word that's flagged for AWB typo correction might actually be correct in certain circumstances. The example I came across recently was that a user did a major AWB change from "cliche" to "cliché" on articles pertaining to a Canadian politician whose name was Robert Cliche, with no accent because it's a French language surname, pronounced "Kleesh", which was not meant to convey that Mr. Cliche was particularly trite or clichéd. Is there any way to flag certain specific articles for an automatic exemption from such spelling "corrections" which would actually be incorrect? Bearcat (talk) 11:10, 15 January 2008 (UTC)

The ruling with the typo list is, either, the page can be excluded completely (only really for certain pages), or, if they produce ANY false positives, they should be removed from the typo list. —Reedy Boy 11:57, 15 January 2008 (UTC)
A possible workaround is to ignore the words in title of the page. — Dispenser 20:26, 9 March 2008 (UTC)

WebControl.Wait()

MaxSem identified an issue in Wikipedia_talk:AutoWikiBrowser/Bugs#Logged_in_user_at_bottom_right_still_not_updating, and User:Skenmy asked me if i could make it autoprotect for Wikinews.

Bit of fiddling, and the code would work.. But the Wait is releasing it too early.. And its going on to load the edit page, causing it not to fully protect

Blasted thing! Any suggestions? Must be another way round it...?

—Reedy Boy 23:04, 24 January 2008 (UTC)

AWB almost functional in Wine!

See last two comments on http://bugs.winehq.org/show_bug.cgi?id=8499 - a workaround in AWB may be possible, though the ideal solution would of course be patches to Wine ;-) In any case, it's now working well enough to spot bugs in! - David Gerard (talk) 13:56, 17 February 2008 (UTC)

I want added to the Project so i can use it my name is Milece and this is the only talk page i found ive clicked through it numerous times, so please add me and i apologize if this is the wrong page. —Preceding unsigned comment added by Milece (talk • contribs) 09:45, 6 March 2008 (UTC)

Could you guys...

add a list reduction mode? So user can create a big list, let AWB eliminate links that don't need attention. After that it would be possible to work only with list that do require user attention for sure, at convenient time. And second - can articles specific spellchecker exceptions be created? I mean let user mark some words at exception belonging to article. And let AWB upload them to some page. That page could be read at start of AWB, and later would let to avoid multiple human intervention and associated mistakes... TestPilot 21:59, 11 March 2008 (UTC)

Feature requests are this way. MaxSem(Han shot first!) 06:11, 12 March 2008 (UTC)

Addition to Custom Project Drop-Down

Hey guys. Recently downloaded the BOT and found that there is no wiki option. I work on http://www.wiki.tribalwars.net currently there are issues with BOT Api and SVN but if there was a wiki option it would save alot of hassle in the future :) --Oscardog1991 (talk) —Preceding comment was added at 13:05, 12 March 2008 (UTC)

Im not sure what your asking for... —Reedy Boy 13:11, 12 March 2008 (UTC)
I don't think that your wiki is large enough to warrant its own variant in project settings. You will need to input your URL only once because AWB saves such settings. MaxSem(Han shot first!) 13:24, 12 March 2008 (UTC)
Yeh but as you know it current doesn't work. Can you reply on the Discussion page of the AWB Bot so i can find out where to update my SVN and BOT Api or explain how to update it etc :) Thanks :) --Oscardog1991 (talk) —Preceding comment was added at 14:46, 12 March 2008 (UTC)
Try the "custom" option, and then inputting www.wiki.tribalwars.net. --76.105.227.156 (talk) 04:07, 14 March 2008 (UTC)

AWB's home

AWB isn't really a Wikipedia EN project any more, although all the devs are of course active here. I've no objection to staying here, but might it not send out a good message if we were to move the AWB infrastructure to a "neutral" wiki, be it meta, mediawiki.org or our own wiki? --kingboyk (talk) 21:19, 9 May 2008 (UTC)

Own wiki isn't as reliable as Wikimedia cluster, and can be easily slashdotted. AWB is offtopic on mw.org, and of all us devs only I have +sysop on Meta, and we often need to tweak a protected page with settings. Yes, "hosting" here is a bit biased, but is more convenient. MaxSem(Han shot first!) 16:53, 18 May 2008 (UTC)

Release next version please

A couple of weeks ago I suggested that it was appropriate to release a new version of AWB, particularly due to the typo list loading bug Wikipedia_talk:AutoWikiBrowser/Archive_18#Version_bump_after_typo_list_access_fix.3F. A user is having problems because of this bug - [2]. I really think we need a new (optional update?) release to roll out this fix. Opinions? Thanks Rjwilmsi (talk)

As the user mentioned above, I concur - "forname" is a common misspelling, and being unable to use the new typo list is a pain (it currently corrects it to "oref") - I've already had at least one bad edit slip through because of it. — iridescent 22:08, 17 May 2008 (UTC)
Agreed. If there will be no objections from other devs, I'll release it in a few hours. MaxSem(Han shot first!) 16:55, 18 May 2008 (UTC)
Released, old version disabled. MaxSem(Han shot first!) 10:25, 19 May 2008 (UTC)