This release only contains some bugfixes, it's pretty minor thing.
I would rather call it 3.1.2.1 but since we don't use these cute long
versions... meh
Hi,
I recently updated code of updater in order to get some better stats
of huggle usage: http://tools.wmflabs.org/huggle/updater/updates.txt
Unfortunately I figured out that tool labs don't provide any way to
collect (even anonymized) hostnames of people who are opening the
website.
For this reason I would like to move the updater out of tool labs to
huggle.wmflabs.org (our private instance on wmflabs that is already
used for a number of services related to huggle, most notably
whitelist and ClueBot relay), so that we could make a distinction in
statistics based on IP address of users who do use it. Note these IP
addresses would be turned to md5 hashes.
Is there any objection or some opinion on this?
Yes, that would be nice to have thing.
Right now there is something I call HAN (huggle antivandalism network)
which is a very simple IRC based service that huggle clients and other
software can use to communicate with each other -
https://en.wikipedia.org/wiki/Wikipedia:Huggle/HAN
this is what we use now to propagate scores that are computed by
ClueBot to all huggle clients and of course we use it to make it
possible for huggle clients communicate with each other :) (there is
even a chat window that can be used by huggle users to talk to each
other within a project scope)
On Fri, Aug 22, 2014 at 4:55 PM, Aaron Halfaker
<aaron.halfaker(a)gmail.com> wrote:
> Thanks Petr,
>
> I'll post to the huggle list next time. For an example score from STiki's
> API, see
>
> http://armstrong.cis.upenn.edu/stiki_api.php?style=score&rid=346523549
>
>
> For more info, see the readme:
> http://armstrong.cis.upenn.edu/stiki_api_readme.txt
>
> I'll be writing up a proposal for a general edit scoring service API. It
> would be nice if that could include a "Huggle" score along side others. I'll
> be digging through your code to see what it would take to extract such a
> score for a common API.
>
> -Aaron
>
>
> On Fri, Aug 22, 2014 at 3:49 PM, Petr Bena <benapetr(a)gmail.com> wrote:
>>
>> Hi,
>>
>> It's always a better idea to post to huggle list instead of me only.
>> For example, now I was on vacation for few weeks and couldn't answer
>> you :-)
>>
>> Yes, it is possible to USE IT, but right now huggle support for non
>> english wikis which are maintained either by some people who don't
>> understand huggle or people who do understand huggle but not the
>> language, isn't very good, so even scoring isn't very good. The very
>> algorithm is pretty simple, you can see it at
>>
>> https://github.com/huggle/huggle3-qt-lx/blob/master/huggle/wikiedit.cpp#L92
>> however, resulting score isn't shared in any way. Only possible way to
>> accomplish this would be to run a huggle based bot that would send the
>> scores somewhere. This is actually possible (I am working on head less
>> huggle that could run as a system service instead of GUI) but it's not
>> up yet...
>>
>> In addition I don't understand how STiki shares the scores? How is that?
>>
>> On Tue, Aug 19, 2014 at 11:46 PM, Aaron Halfaker
>> <aaron.halfaker(a)gmail.com> wrote:
>> > Hi Petr,
>> >
>> > I'm the maintainer of Snuggle[1]. Snuggle uses vandalism scores from
>> > STiki[2] on enwiki to generate desirability scores for newcomers. I'd
>> > like
>> > to extend Snuggle to non-English wikis, but there's no STiki outside
>> > enwiki.
>> > I noticed that Huggle exists outside enwiki *and* it performs a triage
>> > of
>> > edits by the probability that they are vandalism.
>> >
>> > Is there a way that I could make use of Huggle's scoring system in
>> > Snuggle?
>> >
>> > 1. https://en.wikipedia.org/wiki/Wikipedia:Snuggle
>> > 2. https://en.wikipedia.org/wiki/Wikipedia:STiki
>> >
>> > -Aaron
>> >
>> >
>
>
Hi,
It's always a better idea to post to huggle list instead of me only.
For example, now I was on vacation for few weeks and couldn't answer
you :-)
Yes, it is possible to USE IT, but right now huggle support for non
english wikis which are maintained either by some people who don't
understand huggle or people who do understand huggle but not the
language, isn't very good, so even scoring isn't very good. The very
algorithm is pretty simple, you can see it at
https://github.com/huggle/huggle3-qt-lx/blob/master/huggle/wikiedit.cpp#L92
however, resulting score isn't shared in any way. Only possible way to
accomplish this would be to run a huggle based bot that would send the
scores somewhere. This is actually possible (I am working on head less
huggle that could run as a system service instead of GUI) but it's not
up yet...
In addition I don't understand how STiki shares the scores? How is that?
On Tue, Aug 19, 2014 at 11:46 PM, Aaron Halfaker
<aaron.halfaker(a)gmail.com> wrote:
> Hi Petr,
>
> I'm the maintainer of Snuggle[1]. Snuggle uses vandalism scores from
> STiki[2] on enwiki to generate desirability scores for newcomers. I'd like
> to extend Snuggle to non-English wikis, but there's no STiki outside enwiki.
> I noticed that Huggle exists outside enwiki *and* it performs a triage of
> edits by the probability that they are vandalism.
>
> Is there a way that I could make use of Huggle's scoring system in Snuggle?
>
> 1. https://en.wikipedia.org/wiki/Wikipedia:Snuggle
> 2. https://en.wikipedia.org/wiki/Wikipedia:STiki
>
> -Aaron
>
>
Wheeeeeeeeeee,
New huggle was just released, based on low feedback from beta version
I decided that there are no known bugs. :D
Builds for all OS are now being created so don't be surprised if there
is no package for your OS so far.
Hi,
Please test huggle 3.1 beta (windows build at
http://tools.wmflabs.org/huggle/files/huggle3.1.0beta.exe) as we are
going to release it soon. We didn't receive much feedback so far
Major changes:
* Multisite support - you can now log in to unlimited number of wikis
in 1 huggle session and get a huge queue of all edits made to these
wikis. This is good for smaller projects which gets overlooked often.
* Ranged diffs - you can select multiple revisions and get a huge diff
that display all changes done to them.
* Fixes of most of bug reports we had so far
In case you found a bug, please report it to bugzilla:
https://bugzilla.wikimedia.org/buglist.cgi?product=Huggle&list_id=147663
thank you!
I implemented a profiler to latest master, you can compile huggle by
uncomenting PROFILING in definitions.hpp, it is written using C++
macros and its code is replaced with NOOP code in case it's not
enabled.
Example output
Tue Jul 29 17:55:00 2014 DEBUG[1]: PROFILER: Core::Init() finished in 365ms
Tue Jul 29 17:55:00 2014 DEBUG[1]: PROFILER: Login::Login(QWidget
*parent) finished in 527ms
Tue Jul 29 17:55:09 2014 DEBUG[1]: PROFILER:
MainWindow::MainWindow(QWidget *parent)@layout finished in 303ms
Tue Jul 29 17:55:09 2014 DEBUG[1]: PROFILER:
MainWindow::MainWindow(QWidget *parent)@welcome finished in 0ms
Tue Jul 29 17:55:09 2014 DEBUG[1]: PROFILER:
MainWindow::MainWindow(QWidget *parent)@providers finished in 139ms
Tue Jul 29 17:55:09 2014 DEBUG[1]: PROFILER:
MainWindow::MainWindow(QWidget *parent)@layout finished in 54ms
Tue Jul 29 17:55:09 2014 DEBUG[1]: PROFILER:
MainWindow::MainWindow(QWidget *parent)@irc finished in 41ms
Tue Jul 29 17:55:09 2014 DEBUG[1]: PROFILER:
MainWindow::MainWindow(QWidget *parent)@hooks finished in 0ms
So, for example you can see that it took almost half a second for
login form to render (which is quite a lot).
You can insert your own profiler checks using these macros:
HUGGLE_PROFILER_RESET; // will reset internal profiler timer, it's a
good idea to do this before you start profiling some code
HUGGLE_PROFILER_PRINT_TIME ("name of function@id"); // will print a
time it took computer to get from last profiler reset until the mark
to debug log
have fun optimizing huggle!
Hi,
I did an emergency release of huggle 3.0.6 which fixed quite critical
bug that makes the user info widget unusable:
https://en.wikipedia.org/wiki/Wikipedia:Huggle/Feedback#User_Info
The mere fact that nobody noticed this implies all huggle devs are
lame :P and beta testers probably defunct.
I will try to make our beta team more operational as there is a huge
number of testers listed on our members page, but probably nobody beta
testing as this bug was really visible on first sight.