I would suggest to add each request as an individual bug issue,
allowing the person who wants to tackle the issue to work on it
## "SMW is slow" indicator
Adding a query execution duration as meta data to the "has query"
subobject (SMW\QueryData) is not that difficult and could be done by
anyone without specific SMW knowledge. I'm not sure about the "being
slow" indicator but having a query duration meta info would allow to
aggregate a mean value that could be used as temporal indicator about
increased/decreased query durations.
I'd expect Special:SemanticStatistics to be more a dashboard (with
visual aids) then being a repetition of what is already being listed
PS: Both the SMW\ApiInfo and the SMW\StatisticsCollector have
sufficient test coverage and any enhancement should also be applied to
its unit test.
On 7/5/13, Jamie Thingelstad <jamie@...> wrote:
> I'm starting to modify my bots to collect the new information that you put
> in the SMWinfo API calls. I love how you added format count as a
> dictionary/hash. The full information for WikiApiary  looks great. I'm
> particularly pleased that formats from Semantic Results Formats show up as
> Not sure if you know this but I hacked up a "pull JSON from Wikitext
> solution to get this as well" . You can see it's output . About 15-20
> wikis have installed this page and my bot pulls it to get more info. I would
> like to get rid of this, as I don't think long term it will work well and
> it's too cumbersome to implement and support. I also implemented it before
> you made your improvements to SMWInfo, which now gives the format
> information I was looking for, and does it much better.
> Sorry for being overly verbose. My email was to see if you would consider
> adding a couple more things to the SMWInfo request. I think it would be nice
> to have:
> pagequerycount: The number of pages that have a query on them. (this is in
> my semantic usage page) Why? I think this is useful to show how broadly SMW
> is used in the wiki. What percentage of pages have queries?
> querysizescount: This would be similar to the format count, but would show
> the count of queries at each size. This is in my Semantic usage thing as
> well, and I punted on size after 10 and just did anything over 10. Ideally,
> I think showing a dictionary with each used count size would be great. Why?
> I think it's useful to see how complex the queries are that are being
> These two things are the only thing I collect in my hack solution that are
> not currently in SMWInfo. While considering it, I do think it would be nice
> to have:
> typecount: Another one like format count that would output a dictionary.
> This would show the distribution of types (text, boolean, date, number, etc)
> of properties in the wiki.
> querydepthscount: Same as querysizescount but for depth.
> It would also be pretty sweet to bring these new SMWInfo values back into
> Special:SemanticStatistics for people who don't dive into the API.
> I'd be happy to put these into Bugzilla, just let me know. Or, if you think
> they are bad ideas that's cool too… I'm likely going to get rid of SMWUsage
> either way. I also have no real good understanding of how much work this
> stuff is. If I'm asking for something huge, please excuse me since that
> definitely isn't my intent.
> PS: I''m not going to ask for it since the raw data isn't there, but it sure
> would be interesting if the Has query subobjects that SMW creates had a
> field for the time it took to execute the query. I would love to have that
> just for my own wikis so I could keep an eye on very expensive queries.
> Pulling that data somehow would be interesting, but would need to be more
> than just average or something. Probably a second value for my proposed
> querysizescount so it could show that there were 58 queries of size 2, and
> those 58 queries averages 0.07 seconds. Something like that may help with
> the "SMW is slow" beliefs that are out there.
> Jamie Thingelstad
> mobile: 612-810-3699
> find me on AIM Twitter Facebook LinkedIn