On Montag, 6. April 2009, Yaron Koren wrote: > Markus - I think any info on the type of the parameter can just be part of > the parameter description; an example would be, for the "zoom" parameter of > the "googlemap" format, "The zoom level of the map (must be an integer > between 1 and 18, with '1' being the furthest away)".
Indeed, this would be a good start. > > Sergey - it's true that creating a good UI is hard, but I don't think > having each format provide its own forms is the way to go - first because > it's a lot more work, and second because, if we do come up with a good UI, > I bet it'll be fine across all formats. I tend to agree here. The biggest drawback I see with the "class-generated form" option is that you get only one possible form. This is useful for MW prefs, since the UI into which the forms are embedded is known and fairly stable. But I think one of the main motivations of extensions like Halo is to improve the UI in new and fancy ways (e.g. with auto completion). If printers were to ship ready-made forms, then Halo would not have much freedom to design new interfaces, and the task of improving UI functionality would become a responsibility of the printer developers (including me ;-). This is why I would prefer an approach where only the relevant parameters are exported, and the form construction is left to the code using this. I think the following API would be most extensible and convenient: * Have a function that returns an array of parameter names. * Have further functions to retrieve extra information about a particular parameter (the input would be the parameter name). Examples include getParameterLabel(), getParameterDescription(), and maybe, one day, getParameterType(). > > To illustrate my point for both, I put together a mockup of what a simple > interface, using the parameter declarations, could look like. Using > Javascript, the form would change the set of parameters available for the > user to fill out every time they re-selected a format from the dropdown. > Here's what it would look like if the user selected the 'list' format: > > http://discoursedb.org/new-special-ask.png Frankly, the first thing I would change for an improved Special:Ask would be the two topmost text inputs. Especially adding printouts should be easier and more structured. But Special:Ask so far was not meant to be very user friendly, but rather as a web service that "further results" and data exports can link to. I think it would be worth cleaning up the code (Special:Ask is still rather messy) and then consider separating the UI aspects more clearly from the web service aspects, maybe far enough to allow for simple subclassing and customisation of the whole special page. -- Markus > > -Yaron > > > On Mon, Apr 6, 2009 at 10:57 AM, Sergey Chernyshev < > > sergey.chernys...@gmail.com> wrote: > > Hey, I'm back from San Francisco and what I find? the Special:Ask > > discussion is getting hot ;) > > > > I have a few things to input - my experience shows that building > > universal UI builder is a hard thing to do - Yaron can probably second > > that. My approach to this problem before was to have subclasses build > > their own forms and parse their own parameters, similar to how extending > > MW preferences interface is done. > > > > This is a way to avoid new constants and constraints on input UI > > parameters for specific query printers, more over, since all of this > > functionality is implemented in PHP and not reading user-generated > > content within the pages, I think it makes the most sense to just leave > > it to subclassing approach which will allow more flexibility for SRF > > implementors. > > > > Thank you, > > > > Sergey > > > > > > -- > > Sergey Chernyshev > > http://www.sergeychernyshev.com/ > > > > > > On Mon, Apr 6, 2009 at 10:01 AM, Markus Krötzsch < > > > > mar...@semantic-mediawiki.org> wrote: > >> I am convinced :-). What remains unclear to me is how to report the > >> expected > >> input type of a parameter to a UI-creating interface. The types are not > >> really > >> standard things like "integer" and "string" but also things like "some > >> template name" or "the name of a property printout in the query". Should > >> we > >> introduce yet another system of type constants? > >> > >> -- Markus > >> > >> On Montag, 6. April 2009, Yaron Koren wrote: > >> > Hi, > >> > > >> > This is great - the self-description/"reflection" feature will be > >> > allow > >> > >> for > >> > >> > a big improvement to Special:Ask (and evidently Halo as well). This, > >> > in addition to Sergey's previous changes, turns Special:Ask from > >> > somewhat > >> > >> of a > >> > >> > hacker interface to an easy-to-use entry point for those getting > >> > started with Semantic MediaWiki and inline queries. Actually, the more > >> > I think about it, the more I think this is a big step forward in > >> > turning SMW > >> > >> into a > >> > >> > real consumer application, since it will mean that people can run (and > >> > administrators can save) queries on the data without needing to read > >> > through a lot of documentation. > >> > > >> > To that end, I would say that yes, there should be descriptive labels > >> > >> for > >> > >> > each printer, retrieved from message files - just as there should be > >> > >> labels > >> > >> > for each parameter. Having good inline help is very common among > >> > mature software applications - it's considerably easier to use than a > >> > user handbook. Obviously, it becomes an issue if someone wants to put > >> > whole paragraphs or even pages into the application, but a simple > >> > one-sentence description of each concept is, I think, completely > >> > reasonable. And I should note another veery important reason to put > >> > such information into > >> > >> the > >> > >> > application - to enable translation; which should make creating inline > >> > queries quite a bit easier for those who can't speak English. > >> > > >> > Finally, if you're at all considering adding more meta-data to each > >> > printer, let me suggest another field: a "default limit" value, which > >> > >> would > >> > >> > override the wiki's value for $smwgQDefaultLimit (though not a > >> > "limit=" parameter set in the query). This would allow the "math" > >> > formats in SRF (sum, average, min, max) to considerably raise the > >> > number of values > >> > >> passed > >> > >> > in to them by default, since all of them, especially 'sum', require > >> > all values to be passed in to produce a meaningful result. It would > >> > >> similarly > >> > >> > be helpful for 'calendar'. A default limit value could also be helpful > >> > >> for > >> > >> > the export formats (csv, json, icalendar, rss), although for them > >> > >> there's > >> > >> > already a workaround because they create, essentially, a link to > >> > themselves, with that link setting the correct limit. > >> > > >> > -Yaron > >> > > >> > > >> > On Sun, Apr 5, 2009 at 12:24 PM, Markus Krötzsch < > >> > > >> > mar...@semantic-mediawiki.org> wrote: > >> > > As promised, SMW has now been updated to support some of the > >> > > requested functionalities: > >> > > > >> > > * To get the list of all available formats, access > >> > > $smwgResultFormats. * To get a result printer object, use > >> > > SMWQueryProcessor::getResultPrinter(). > >> > > * To get the name of a printer object, use $printer->getName(). > >> > > > >> > > The new functions are documented in the code (or see the API docs). > >> > > > >> > > The format selector for Special:Ask has been improved to use these > >> > > features. > >> > > Note that not all SRF formats provide localised names yet; examples > >> > > on how to > >> > > add translations are found in all SMW result printers and in some of > >> > >> the > >> > >> > > SRF > >> > > printers. > >> > > > >> > > There was also another change: printers that only show download > >> > > links when used inline can indicate to SMW that no query needs to be > >> > >> executed > >> > >> > > in this case. To do this, simply implement the method getQueryMode() > >> > >> as > >> > >> > > e.g. in SMW_QP_CSV.php. If you do not do this, SMW will always > >> > > compute all query results, even if the printer ignores them anyway. > >> > > > >> > > Besides Name and QueryMode, no further meta-data has been added. > >> > > Especially, I > >> > > do not know yet how to provide information about available > >> > > parameters. There > >> > > is a list of parameter identifiers, but what else should be there > >> > > for each of > >> > > them: localised labels, expected type of input, anything else? Also, > >> > > I was unsure whether longer descriptions besides the names should be > >> > > available for > >> > > printers -- doing this feels like moving the user handbook to the > >> > >> message > >> > >> > > file. > >> > > > >> > > Cheers, > >> > > > >> > > Markus > >> > > > >> > > On Donnerstag, 2. April 2009, Markus Krötzsch wrote: > >> > > > On Mittwoch, 1. April 2009, Yaron Koren wrote: > >> > > > > I think a way for formats to publish both a description of > >> > >> themselves > >> > >> > > and > >> > > > >> > > > > the set of parameters they take would be very helpful: it could > >> > > > > be that the easiest way to do it is just to expand the > >> > > > > information > >> > >> that > >> > >> > > > > $smwgResultFormats holds, although maybe the better solution is, > >> > >> as > >> > >> > > > > you suggest, to add methods like getDescription() and > >> > > > > getAllowedParams() to the SMWResultPrinter class, and allow any > >> > > > > format class to override > >> > > > >> > > these > >> > > > >> > > > > methods. > >> > > > > >> > > > I also agree with Jörg's observation. I strongly favour Yaron's > >> > >> second > >> > >> > > > proposal: extending the printer class instead of extending the > >> > >> static > >> > >> > > > array. A related solution has been in my mind for some time, but I > >> > >> did > >> > >> > > not > >> > > > >> > > > find time to implement it yet: > >> > > > > >> > > > * In order to keep the information that is available for each > >> > > > result printer flexible and extendible, it would be best if the > >> > > > printer objects would implement access methods for this > >> > > > information. Two > >> > >> main > >> > >> > > > advantages are: > >> > > > > >> > > > (1) The meta information is kept with the code of the printer > >> > > > class, > >> > >> so > >> > >> > > > printer format extensions like SRF are still easy to implement in > >> > >> one > >> > >> > > file. > >> > > > >> > > > (2) The abstract result printer parent class can always implement > >> > > > >> > > fallback > >> > > > >> > > > methods, so that non-core printers are compatible to new SMW > >> > >> versions > >> > >> > > even > >> > > > >> > > > if they do not support this meta-data yet. > >> > > > > >> > > > * The task then boils down to getting a printer object for a given > >> > > > >> > > printer > >> > > > >> > > > identifier (like "ul"). In order to be perfectly compatible with > >> > >> SMW, > >> > >> > > this > >> > > > >> > > > should be done by calling SMWQueryProcessor::getResultPrinter(). > >> > > > The problem is that this function currently requires further > >> > > > parameters: especially it depends on the query result. This is not > >> > > > really needed and can be avoided with some extra coding from my > >> > > > side. > >> > > > > >> > > > * What remains to discuss is what meta information it is that each > >> > > > >> > > printer > >> > > > >> > > > should provide. A human readable name would be a good start, I > >> > >> guess. > >> > >> > > > Another useful feature might be a function that returns the list > >> > > > of supported additional parameters with a short "signature" to > >> > > > indicate > >> > > > >> > > their > >> > > > >> > > > input type (How should this look? How do you get > >> > > > labels/descriptions for these parameters?). Anything else? > >> > > > > >> > > > I think I can provide the implementation over the weekend then. > >> > > > > >> > > > -- Markus > >> > > > > >> > > > > -Yaron > >> > > > > > >> > > > > On Wed, Apr 1, 2009 at 4:25 AM, Jörg Heizmann < > >> > >> heizm...@ontoprise.de> > >> > >> > > wrote: > >> > > > > > Hi, > >> > > > > > > >> > > > > > I am working on our query front-end for inline ask queries > >> > >> which, > >> > >> > > > > > in the next version, immediately shows your result set while > >> > > > > > you are creating your query. > >> > > > > > In order to (dynamically) support all available result > >> > > > > > printers > >> > >> and > >> > >> > > to > >> > > > >> > > > > > not hardcode all this stuff which might change in the future I > >> > > > > > found the array holding name and class of each available > >> > > > > > result printer ($smwgResultFormats). That's pretty cool but I > >> > > > > > cannot > >> > >> get > >> > >> > > > > > additional meta information for each printer. > >> > > > > > > >> > > > > > I doubt that users understand that "ul" stands for "unordered > >> > >> list" > >> > >> > > so > >> > > > >> > > > > > I would love to have a short description (maybe also a longer > >> > >> one) > >> > >> > > > > > method for each result printer. > >> > > > > > Also, some printers support additional parameters which are > >> > > > > > necessary for the frontend to know about. Here, a second > >> > > > > > method would be nice > >> > > > >> > > to > >> > > > >> > > > > > have in order to not hard-code this information. > >> > > > > > > >> > > > > > What do you think about these additions? > >> > > > > > > >> > > > > > cheers, > >> > > > > > Jörg > >> > >> ----------------------------------------------------------------------- > >> > >> > > > > >-- ----- _______________________________________________ > >> > > > > > Semediawiki-devel mailing list > >> > > > > > Semediawiki-devel@lists.sourceforge.net > >> > > > > > https://lists.sourceforge.net/lists/listinfo/semediawiki-devel > >> > > > >> > > -- > >> > > Markus Krötzsch > >> > > Semantic MediaWiki http://semantic-mediawiki.org > >> > > http://korrekt.org mar...@semantic-mediawiki.org > >> > >> -- > >> Markus Krötzsch > >> Semantic MediaWiki http://semantic-mediawiki.org > >> http://korrekt.org mar...@semantic-mediawiki.org > >> > >> > >> > >> ------------------------------------------------------------------------ > >>------ > >> > >> _______________________________________________ > >> Semediawiki-devel mailing list > >> Semediawiki-devel@lists.sourceforge.net > >> https://lists.sourceforge.net/lists/listinfo/semediawiki-devel -- Markus Krötzsch Semantic MediaWiki http://semantic-mediawiki.org http://korrekt.org mar...@semantic-mediawiki.org
signature.asc
Description: This is a digitally signed message part.
------------------------------------------------------------------------------ This SF.net email is sponsored by: High Quality Requirements in a Collaborative Environment. Download a free trial of Rational Requirements Composer Now! http://p.sf.net/sfu/www-ibm-com
_______________________________________________ Semediawiki-devel mailing list Semediawiki-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/semediawiki-devel
