-
Notifications
You must be signed in to change notification settings - Fork 51
To Do list
Sebastian Humenda edited this page Aug 28, 2016
·
6 revisions
An informal and messy list of things to do. May be used for quick notes before they are ripe enough to be entered into the project tracker
- MFAQ: add the non-server method of access (Open Dict)
- modify the maintainer list (see the maintainers issue)
- add information on using SVN (HOWTO, FAQ)
- how about upgrading Michael's freedict-editor to a simple form of TEI-P5?
- Beata and Piotr use FLEx for the non-FreeDict Swahili-Polish dictionary; Piotr plans to create a LIFT-TEI roundtripping system that could be used for FreeDict as well
- project
- introduce changes for OLAC
- restrict the ODD further (hmm, create a higher level ODD, that is... enumerate @type for cit ("dicteg", "trans"?), types of notes(?), etc.)
- think of setting up an ontology for grammatical stuff, either GOLD or (perhaps easier?) ISOCat (see a proposal)
- look at the TEI2stardict stuff for Sebastian (done: I don't think I want to continue talking to them, let someone else take over -PB)
- keep bugging Michael about switching to XSLT 2.0...
- implement symlinks to schemas and such at last (?) -- but some software (oXygen) doesn't understand symlinks (post in oXygen forums); this is important, because otherwise we'll have to update stuff by hand from shared/
- post at least some of this list to the tracker...
- don't forget about the idea of getting rid of the "Maintainer" field from dictionary headers
- rename the non-ISO-named packages (consulting Kęstutis about the timing, do all of them at the same time, when it's convenient for him to grab the renamed packages for packaging) [being done]
- ask Kevin again whether to deprecate or rename wel (whether Kevin's cym is independent and should be versioned separately) --> Kevin suggests deprecation [ok, deleting them]
- modify gender symbols to be more readable than e.g. ( seems better), e.g. in Khasi-English
notes written during the switch to TEI-P5
- use the xml:lang codes: for the source language, target language and the metalanguage
- introduce the header plug-in to the ontology interface
- there are still brackets used in some of them (kha-deu, but also others)
- some equivalents are back to the stage from before Michael's second, rolled-back Ergane import: they are lacking spaces: use the hidden databases to straighten out the equivalents (that is surely not a violation of their license, whatever it is now)
- in those databases, sometimes multi-word entry words are trimmed -- make sure to fix that, too
- talk to Ergane about re-licensing of either the old databases to us (so that they can be un-rolled-back [unrollbacked ? :-)]) or their current databases
- some equivalents are repeated
- some English pronunciation info is suboptimal or just wrong; stress marks are missing, too; possibly, use the English-Polish dictionary as the template, or go for something else that is standard.
- the distribution of equivalents across senses or just cits within a single sense may sometimes be odd (this is a general lexicographic issue)
- possible task: sensible reversal of an existing monodirectional dictionary
- possible task: concatenation of existing dictionaries; it would be good to have a system that may allow for the product of such concatenation to be updated
- necessary task: go over the FreeDict build system, making it absolutely transparent and as modular as possible
- sensible task: use a good dictionary for the basis of pronunciation information (e.g. the English-Polish dictionary is a good beginning for English pronunciation) and project that to all other dictionaries with the same source language
- another sensible task: use the info on equivalent grammatical properties (e.g. gender in *-deu or *-fra dictionaries) and project that to all the others
- style sheets do not split multiple orth's into multiple dict entries (and hence users can't search for them)
These came up in the process of introducing references to externally defined data categories.
- redundancy between n and @ana="#n" -- should we address it somehow?
- it doesn't seem like it can be addressed at tagUsage level
- how to ensure equivalence between vt and vtrans (neither repository has "transitivity" as a general category (GOLD's transitivity is different))
- which of the three "masculine"s (etc.) to choose from ISOCat? (weird)
- ensure that the masc gender in entry/gramGrp (and rel/gramGrp) need not be the same masc as that used in cit/gramGrp -- because they may very well realise different networks of oppositions.