{22} Trac tickets (2647 matches)

e.g. searching for 'buddhist' or 'sanskrit', you don't get this newly created package: ​http://ckan.net/package/digitalsanskritbuddhistcanon

So that Drupal doesn't need to make any adjustments to the form, and risk data being lost on submission.

Move from hetzner box to s3.

ww 22/9/10:

(discussion of ticket:509) This leads us to deduping which should be the next order of business oin this thread...

ww 20/10/10:

if a dataset is missing a department and the ons script makes a new version of it with a department we now have two reccords for the same dataset. these should be manually merged - need a merge tool where you can tell it that two datasets are the same.

Administrator wants to upgrade CKAN or move it to another server. During this time the database is being administered and either edits are lost or can't be done.

We're getting this warning a great deal on live servers. Is these really a sign of the system not operating correctly or can we reduce the level to an INFO?

e.g. on hmg.ckan.net:

2010-10-29 17:12:08,262 WARNI [ckan.lib.search.common] NOOP Index: id,package_id,url,format,description,hash,position
2010-10-29 17:12:08,333 WARNI [ckan.lib.search.common] NOOP Index: id,package_id,url,format,description,hash,position
2010-10-29 17:12:08,375 WARNI [ckan.lib.search.common] NOOP Index: id,package_id,url,format,description,hash,position
2010-10-29 17:12:08,406 WARNI [ckan.lib.search.common] NOOP Index: id,package_id,url,format,description,hash,position
2010-10-29 17:12:08,480 WARNI [ckan.lib.search.common] NOOP Index: id,package_id,url,format,description,hash,position
2010-10-29 17:12:08,613 WARNI [ckan.lib.search.common] NOOP Index: id,package_id,url,format,description,hash,position

As a system developer, I want to run 'create next migration version' script, so that I don't have to create migration version tests and files manually.

The documentation for setting up Apache uses sites-enabled instead of sites-available and a2ensite demo.ckan.net

Data importers should use API instead of object model directly:

• saves load on server
• much better error checking and authentication
• eating our own dogfood (tests the API well)

Intermittent failure on ckan default branch. Looks like brittle test.

======================================================================
FAIL: ckan.tests.models.test_package.TestPackageRevisions.test_02_metadata_created_and_modified
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/buildslave/okfn/full/build/buildandsmoke/pyenv/lib/python2.6/site-packages/nose/case.py", line 186, in runTest
    self.test(*self.arg)
  File "/home/buildslave/okfn/full/build/buildandsmoke/pyenv/src/ckan/ckan/tests/models/test_package.py", line 250, in test_02_metadata_created_and_modified
    assert out == exp, (out, exp)
AssertionError: (datetime.datetime(2010, 11, 25, 4, 37, 52, 949380), datetime.datetime(2010, 11, 25, 4, 37, 52, 94938))

​http://buildbot.okfn.org/builders/builder-ckan/builds/589/steps/shell/logs/stdio

• perhaps we should show the openid as well to distinguish between users with the same name.

• perhaps on account creation, the user should be redirected to their personal details page to encourage them to fill in a human readable name.

• also the list is much too long. can we make it work some other way? javascript is an option, but must be careful to fail gracefully when the browser does not support it.

The CSS @import directive at the bottom of the main CKAN css file is not understood in all browsers, move this to a seperate import.

Test coverage for ckan.lib.cache is only 40%. Can that be usefully improved?

Test coverage for ckan.lib.diff is 24%. Need to improve

​http://bitbucket.org/ww/ckanrdf/changeset/63b60d3a253d

The tests currently take 41 mins to run on most laptops. This slows down development and discourages a test-driven approach.

We'd like to see the tests pass in 5 mins or less (but anything would be an improvement!)

Some suggestions for achieving this include:

• Upgrading the entire codebase to SQLAlchemy 0.6 so that tests could run against an in-memory SQLite database
• Not setting up and tearing down the database so frequently.

Solutions: ResourceSeriesLoader? allows synonyms for values.

I am having trouble installing Ckan with virtualenv and pip using pip-requirements.txt due to the Authentication requirements to download the source code.

Based on the ideas in:

• DistributingChanges
• SyncingInstances

Merging the tickets #296, #297, #298, #299 into this one.

Related ideas include:

• Using dcat to exchange data
• Using couchDB to have a stub package record synchronised between multiple CKAN instances, each linking to the main record in the correct instance

This should be considered a wishlist item for the timebeing.

Every change to every object is being stored in memory, which could add up to quite a lot of memory.

This fixes it by making sure the objects are in a weakref. ​https://bitbucket.org/kindly/vdm/changeset/8d5f91db641f

There are still some intermittent errors due to objects not having a revision_id, if you save an object at the same time as the revision. We add the uuid earlier to fix this. ​https://bitbucket.org/kindly/vdm/changeset/0050d023ca4e.

From David Raznick

David Raznik: I propose we remove autoflush by default in ckan. It caused a lot of bugs on the version upgrade due to its magical behavior.

I have a patch in ​https://bitbucket.org/kindly/ckan/ on branch feature_switch_autoflush_off_by_default.

This change simplifies code. It only took 7 extra flushes/commits to make work and there are over 40 instances that we remove autoflush, so as not to cause error (I have not removed them in the patch yet).

Things to know about using flush:

• Most of the time nothing will change. A commit implies flush.
• You only need to flush if you expect a query to return an object you just saved. i.e if you session.add(obj) and you expect obj to be in a session.query.
• A flush is handy when you want to get the primary key out before you do a commit. i.e session.add(obj); session.flush(); obj.id is now generated.
• You can use them as a limited form of nested transaction.

It increased the performance on the tests by about 5%. There are less flushes altogether so is faster.

I do not see any downsides.

​http://ckan.net/tag/linguistsic doesn't have a single entry, but is still kept and displayed.

It should be purged !!!

Why?

• Will make it easier to give resources their own form, independent of packages.
• Makes uploading, harvesting and previewing data cleaner as its separate from the packages.
• Gives the option of sharing a resource between packages.
• Leads the way to supporting storing richer semantic information against them.

On a production server we produce literally millions of little files saying how long every single request took to process.

They are here:

hmg.ckan.net.2/pylonsdata/call_timing

They were added by ckan/lib/base.py in cset:da438a9085d3.

I don't believe anyone uses these stats.

An internal server error was reported when trying to create a new package (not when editing):

WebApp? Error: <class 'sqlalchemy.exceptions.IntegrityError?'>: (IntegrityError?) null value in column "name" violates not-null constraint 'INSERT INTO package (id, name, title, version, url, author, author_email, maintainer, maintainer_email, notes, license_id, state, revision_id) VALUES (%(id)s, %(name)s, %(title)s, %(version)s, %(url)s, %(author)s, %(author_email)s, %(maintainer)s, %(maintainer_email)s, %(notes)s, %(license_id)s, %(state)s, %(revision_id)s)' {'maintainer': None, 'name': None, 'author': None, 'author_email': None, 'notes': None, 'title': None, 'maintainer_email': None, 'url': None, 'state': 'active', 'version': None, 'license_id': None, 'revision_id': '0141c09c-4eb3-4dda-a6b1-e07f2c73b4c5', 'id': '7fe16a29-f183-471a-99c5-2c2e81dd86e4'}

The error could not be reproduced in a non-apache/mod_wsgi environment, even with a dump of the live database. Further investigation by manual editing of the package form left the "groups" field as the likely cause. After updating the packages installed, and temporarily editing the package controller the problem vanished after a restart of the web server.

Possible causes include:

• Stale pyc file
• Threading issue
• Dependency issue

As the problem is gone for now, it cannot be further investigated atm.

At the moment the assignment of user roles on package creation is done in setup_user_roles method in ckan/model/authz.py and is hard-coded.

This can be a pain to override (you can use a Plugin listening for package create events) and makes some things such as putting system in restricted mode more complex than it should be (see ticket:833).

An elegant solution would be to move this into an Extension this simplifies the code and make it easier for people override (just remove the default extension and plugin your own).

The new field was outlined in

​http://lists.okfn.org/pipermail/ckan-discuss/2011-February/000887.html

it says

Can be "data", "metadata", "service", NULL. There is a need to do this for UKLP where some of the data associated with a package is metadata, some is data and some could be a service.

We are now harvesting metadata from other sources in various places around CKAN. Such harvesting can include:

• CSW/WFS for INSPIRE/UKLII (yields CKAN packages)
• Catalogue scraping for LOD2 experiments (yields RDF graphs)
• Atom/DCat for LOD2 production (yields RDF graphs)
• OAI-PMH for ​http://datadryad.org/ and other dspace (yields CKAN packages)

We should aim to consolidate the harvesting clients into a common system that is easy to extend when needed and can be re-used in different scenarios.

In general, such a system would have the following stages:

• Source selection: find what to download/scrape/harvest/parse
• Index retrieval (i.e. package index)
• Item retrieval (i.e. package entity)
• (Optional: Serialization)
• Normalisation
• Loading/Merging? into CKAN

Exisiting harvesters are at:

• CSW: ​https://bitbucket.org/okfn/ckanext-csw/src/
• Scraper+CKAN: ​https://bitbucket.org/pudo/dcat-tools/src/d5d96b06ec9a/dcat/crawl/

Try and minimise the amount of different caching methodologies, especially for api calls.

When doing create on the cli upgrade should be run instead of create to make sure that every database is initiated in the same way.

Needed to allow extensions to be internationalized in the code and in templates. Requires in-template i18n domain switching which is broken between Genshi and Pylons, thus needs this hackish class.

cf. ​https://bitbucket.org/okfn/wdmmg/src/198841e770e5/wdmmg/config/environment.py#cl-16

add revisioning to harvested document table

1. Login (shows ckan home page with flash message "Welcome back xyz")
2. Click away, to "Add package" say
3. Click back to ckan "Home". It shows "Welcome back xyz" again - it shouldn't, I assume?

It's not good to have ckanext doing lots of different things with different dependencies. Split it off into:

• ckanext-importlib

etc.

And then deprecate the ckanext repo itself.

The problem is that in ckan/controllers/rest.py the BaseApiController? has this method:

    def __before__(self, action, **env):
        BaseController.__before__(self, action, **env)
        if not self.authorizer.am_authorized(c, model.Action.SITE_READ, model.System):
            abort(401, _('Not authorized to see this page'))

which works on the basis of your c.user, rather than your apikey. All API users are treated as visitors (since API users don't get a login cookie) and even a sysadmin's apikey is blocked unless there is a right for a Visitor to SITE_READ.

Also needs tests.

(Also, why is this restriction only on the API, package search, group index and tags and agroup index? I'm guessing SITE_READ is only for places where other authz don't apply, but maybe it should not be called 'SITE_READ' but 'OTHER_READ' or something?)

Refactor ckan/lib/authztool.py so that the relevant methods are independent of the command line interface.

The extracted methods should live in a new file ckan/authz.py. authztool.py should probably move into cli.py and will just do command line parsing and printing and use ckan/authz.py. The updated web gui for authz will also use this code.

Tests should be made. There's already a file ckan/tests/test_authz.py, which looks like the appropriate place for new tests.

all to go on a branch feature-1050-refactor-authtoolz

Optional extras

• Rename ckan/authz.py to ckan/lib/authz.py or even ckan/logic/authz.py

I've been asked if we can do something about the overflow of the Group name in the right hand column on this page:

​http://register.data.overheid.nl/package/europese-aanbestedingen

The reason is that the list display for groups is in the form "group_tltie (group_name)", and of course group_name can't have spaces and so can't wrap nicely.

I was wondering if there's a good reason why we don't only display group_title (if it exists) and group_name only when there's not a title?

Idea is to have stubs for datasets that someone wants but don't yet exist (or haven't been discovered) in the way one has stub pages on a wiki.

We could do this within the existing model by a slight 'abuse' - create a dataset and mark it with a special tag e.g. todo.does-not-yet-exist or similar ...

(Just as we have datasets listed that exist but aren't available ...)

Alternative would be to have a request for datasets subsystem.

I prefer the stub dataset model because it's simpler, provides a simple workflow (as a dataset is found or comes into existence), and the package page provides a natural space in which to accumulate information about what is wanted and what exists.

Implementation

• Agree a new dedicated tag. e.g. todo.does-not-exist

Related

• This ticket has clear links to ​http://getthedata.org/

Set language to Greek, flash message says 'Language set to: English', but page is now about half in Greek.

Set language back to English causes server error:

AttributeError?: 'NoneType?' object has no attribute 'path'

Module ckan.controllers.error:29 in document view

if original_request.path.startswith('/api'):

However going to a new page reveals that it's back to English

as title.