The jsondata column in the csv file downloadable for index pages is its Python str() and therefore not valid JSON (e.g. None or the u in Python 2 unicode strings):

{u'endangerment': None, u'med': {u'year': 1990, u'doctype': u'grammarsketch', u'id': u'79160',

In order to provide metadata for OLAC, clld apps should support OAI-PMH.

While we'd rather upgrade to bootstrap 3, upgrading from 2.2 to 2.3 does still make sense if only for jquery 1.9 compatibility.

See http://next.datatables.net/blog/2014-02-04 for details. The new version of DataTables provides easier integration with bootstrap, but only version 3 ...

See http://www.datatables.net/download/index

Based on http://programminghistorian.org/lessons/googlemaps-googleearth we should have a tutorial describing how data from various clld apps can be overlaid and visualized on a map.

URLs should look as follows:

/languages
/languages/{id}

i.e. the URL for a single resource should be the URL of the resource class plus the resource id as additional path component.

see http://leafletjs.com/2013/11/18/leaflet-0-7-released-plans-for-future.html for details.

E.g. in the case of Glottolog resourcemap.json?rsc=language should contain about 22 000 resources (active languoids). It only returns about 16 000 (inner instead of outer join in the db query).

Now that webhelpers2 supports python3, our fork of webhelpers.html could be replaced.
(See also http://webhelpers2.readthedocs.org/)

A new release of DataTables is available, including 11 bug fixes, see http://datatables.net/blog/2014-10-09

Clld generates redundant statements of the form $glottocode owl:sameAs $glottocode . in the RDF for resources identified by a glottocode, e.g. languoids in glottolog.

Can this be prevented by special casing?

Since the iconselect is implemented as clickover, it's width is set globally. Thus content has to be arranged in a way, that fits this width and employs vertical scrolling if necessary.

Once we are at bootstrap 2.3, we can upgrade jquery to the 1.9 series.

You might consider semantic versioning http://semver.org.

As described in csv on the web the updated mechanisms for locating metadata should be supported (in addition to the old ones?). Check what's supported by csvlint.io!

The limit on the number of exportable items for the csv adapter should be included in the adapters docstring so that it will be shown to users in the download info popover.

• In the main app template
• in setup.py: newrelic dependency

The root_package attribute added to Pyramid 1.6 might allow us, to use the Configurator object more in line with regular pyramid apps, i.e. clld should be a package that can simply be included, and not one that does create the Configurator object for an app.

See http://www.postgresql.org/docs/9.1/static/functions-matching.html
but also the syntax recognized by clld.db.util.icontains.

The clld package should be independent of the CLLD project's publishing infrastructure. So

• adapt default app footer,
• remove fabfile from app scaffold,
• remove legal.mako,
• remove default values for dataset objects.

A scaffold for a clld app to be part of CLLD should be provided in a separate package.

Leaflet 0.7.3 is the stable version as of May 23, 2014 (see http://datatables.net/manual/styling/classes).

Custom resources so far require

• an interface and a model,
• a template directory, including an rdf.mako template,
• a DataTable.

Some (most?) of this should have usable defaults.

The two functions to make common clld functionality available for clld apps should be merged. The pkg argument to register_app is not necessary because the calling package can be accessed via the Configurator's package (and package_name) attribute(s).

A (configurable) API should be implemented, exposing domain-specific operations, like

• updating language data (ISO code, ...)
• re-coding values

in a way that makes it easy to put a JavaScript UI on top.

Authentication should be really simple, i.e. just put URL path behind HTTP(S?) Basic Auth.

The API for reading any kind of delimiter-separated values should be unified, preferably following the API of the standard lib csv.reader.

Could be implemented using https://github.com/cleder/fastkml

Note: This requires a lot of changes, in clld as well as in basically all non-trivial clld-apps.

• upgrade to jquery 1.10.x
• upgrade DataTables (see https://github.com/DataTables/Plugins/tree/master/integration/bootstrap/3)

Right now, CSV views of data tables put the columns in alphabetic order.

How about producing them in the order of the model fields? E.g. like this:

def csv_head(self):
    mappers = list(object_mapper(self).iterate_to_root())
    exclude = {'active', 'version', 'created', 'updated', 'polymorphic_type'}
    cols = [col.key for om in reversed(mappers) for col in om.local_table.c
        if col.key not in exclude and not exclude.add(col.key)]
    return cols

We currently pin html5lib at version 0.999, but recent rdflib releases require 0.95. Figure out how to resolve this in a way that allows us to work with most recent versions of major packages.

The processing message does only appear once when the table is initially loaded, but not on any further events. The following jsfiddle may help resolve the issue (possibly changing the visibility manually): http://jsfiddle.net/Ajinkya_Parakh/TNe37/5/light/

deployment of apps in production mode needs to be supported:

• log rotation
• deploy under separate domain served via nginx site

All functionality dependent on fabric should be factored out into a new package to

• make migration to python3 easier,
• separate deployment/clld project specific stuff from the community package.

These fonts are useful for creating PDF documents containing unicode text, which is done in the csd and tsammalex apps already.

Deployment of a clld app should encompass creation of an associated solr core.

create temporary parameter with combined domain?