# Some ideas and notes Most of these ideas may be a lot more general, and not really specific to smaoahpa. ### Things that must be done: * Import scripts importing v/V for PoS, needs to be normalized somewhere in import; currently handling uncertainty in other places with iexact * BUG: Change case after receiving results/successes in morfa results in 404 error * Behavior of forms are a little bit funky; pressing enter creates a new set instead of submitting work. Unintuitive. Morfa things not working: * Case selection menu * Contracted stems commented out for now because none exist yet Leksa things not working: * Level selection menu * Language pair selection (commented out everything except sma and nob for now) ### Things that should be fairly easy to do * Sessions can be used to track user progress, such that users who haven't cleared browser cookies can return to see how they're doing * Various repeated actions such as reloading word lists can be done in the admin interface, maybe just add a model for data files containing a path (or allow upload?); and then use admin actions to refresh xml data. - Useful, because the people who maintain the data can then update the data in oahpa as they work on XML, without relying on us to do so. We could just give these permissions in the existing admin interface. + Allowing users to update may require validating data before it is added to DB. + Also need to test whether existing entries are updated or dumped or what.. If we're logging results of users' learning, destroying database entries may screw this up. * Hard-coded lists can be shifted into the DB, e.g., semantic sets available in leksa, but there should be an easy option for populating the database with these things; perhaps fixtures, yaml or some default XML file. - smadrill.views.Gameview - smadrill.forms ### Client requests? ### Things I really want to do * Improved logging. * A lot of the models for word translations are redundant and can be generic. Currently the code dictates what language pairs are available in the system, and this should not be so. * Improve the design/CSS just a bit, more sans-serif, slightly larger fonts * Where are the slow parts in the install/data loading scripts? * Take a look at localization/i18n. Are there other i18n packages available, and can they handle less common languages better? There seem to be problems setting up locales for sma, sme, etc., because the admin interface must have a locale for these languages first. Is it possible to split the user-interface i18n from the admin i18n? * Perhaps a more AJAX-y test method can be developed to make it easier to continually test on words. - A more API-like approach means iPhone/Android apps. And the same API could be used in the web version. - Since this is more of a for-users-own-knowledge/benefit thing, and nothing that requires detailed auditing of client responses, it could be fairly easy to produce something like this; and just send question/answer data. The only place this might get more complex is morphological quizzes and syntax quizzes, since we can't quite expect CG, or an easy way to interact with it to be available on the client side. ;) ### Fun things - Export data to Quizlet? - # Meetings ## Problems/todos identified on thursday TODO: sometimes, however quite seldom, one gets the index out of bounds exception that we had yesterday as the main problem (me and Børre got it each one time) @cip_20101110: this bug is still there! TODO: the bryllupsgabe-bug is not yet fixed: Sjur got @cip_20101110: this bug I haven't got yet, but that doesn't mean that it totally dissappeared! kïeredh ___________________ om rein which is the same type of bug. 3. what is this? baenie hund tann Hunting the bug: sma->nob leekedidh slakte (correct) db says Wordnob w/ slakte as lemma has words in its related set that are leekedidh and leekedh, although according to xml leekedidh is slakte flere dyr... Perhaps this is a reasonable answer. looking at leekedidh, its translations2nob set contains 'slakte', 'slakte flere dyr' and 'flere'. It appears somehow the db adding script is producing some funky additions. ... continuing searching for more eroneous answers. another issue with moehrele tantebarn til mors yngre søster XML says this should be correct, response is wrong. moehrele has two words in nobsme xml. hmm >>> Word.objects.filter(lemma='moehrele')[0].translations2nob.all() [, , ] >>> Word.objects.filter(lemma='moehrele')[1].translations2nob.all() [] >>> Word.objects.filter(lemma='moehrele')[2].translations2nob.all() [] took a look at some more words, and it seems like the database increasingly has more Word (model) words which have duplicate lemmas. I wonder if the install script is having issues adding words when we run the translation... I know it's supposed to be using get_or_create, but it may just not be finding the words properly. Added some testing data in testing/bryllups*, since db was getting 'extra' data. TODO: background and css for the main site TODO: order of icons for plays on the main site sould be the same as in each individual site TODO: compiling and installing the localization anew * BUG: missing lemma sometimes in leksa nobsme: @cip_20101110; this but is still there even after the proper converting of nobsma data. This is not a matter of wrong data format in the XML files but in either smanob data db feeding (the first time) or the nobsma db feeding (on the existing smanob data in the db) or generating random sets from the db when playing. (cf. my two emails wrt. that to Tron, Lene, and Ryan). ad. EL-bug in Leksa nobsma: Debuggin smaoahpa after lang attribute correction in the data. Still found it with ___________________ åahpanadtedh (and now on victorio with ___________________ mojhtelidh ) TODO: Compare these two entries, too. Collecting data sets: in smanob/scr/v_smanob.xml åahpanadtedh bli kjent med flere ....................... in data_sma/sma/v_smanob.xml åahpanadtedh bli kjent med flere ....................... in data_sma/nob/v_nobsma.xml bli kjent åahpanadtedh åahpenidh åahpenadtedh ....................... Todo: check with the database sets --> done! - faulty db feeding - nob entries with no lemma and no pos - re and te elements fed als own entries state-of-art: done after Ryan's fix? To be checked properly! ....................... ....................... General observation: To little time dedicated to testing the data! The workflow is better than that for the first Sma-Nob dictionary in Mac and StarDict format, but still faaaar, faaaar away from what Ciprian expected. Reasons will be discussed in an internal meeting (CLT). TODO (Cip): Write collations and revert sorting (for sma -- Trond's special wish, Kildin -- Michael's special wish) for the xsl scripts. ## 12/11/2010 Todos * db first - morfa: verb class in tooltip - db loading script needs to fix this - models - cases, books: leksa -- js exclusion GENERAL TODO: Get log up and running LEKSA BUG: Selecting a semantic class does not reset book field to 'all'. Intended? MORFA-S VERBS TODO: Pronouns TODO: Present, preterite. TODO: Personal pronouns - ped/sma/doc/smadoc/pronouns_for_morfa_v.csv - See smeoahpa for how pronouns are stored... Are they? or is it just a dictionary in forms? TODO: Click verb, get dictinoary translation, and verb class Klasse IV. (Norsk: vite) - DB loading script bisyllabic - lemma trisyllabic - lemma - Model field NUMRA TODO: The 30 character limit must be tested TODO: The cardinals should be added (now only ordinals) NICETIES ? Add wordcount to each category in Leksa and Morfa? SERVER - Explore WSGI implementation with apache somewhere else, figure out what the documentation means. _________________________________________________ Januar 2011 Røros: Observation, bugs, todo _________________________________________________ tjåegkedh: plukke is NOT accepted, why?