{23} Trac comments (3729 matches)

Results (3201 - 3300 of 3729)

← 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 →
Ticket Posixtime Author Newvalue
#1284 1315212258000000 dread Fixed by rgrp in cset:1952445d2802 on release-v1.4.3 and default. Comment: do not create revision for changes to users as they are not revisioned. * This had already been removed in cset:8d6fde0e2196 (then in user controller) but Adria seems to have reinstated this when refactoring things into the logic layer. This 'bug' explains the huge number of empty revisions we have been seeing on ckan.net.
#1284 1315212603000000 dread I've added tests for this (to prevent yet another regression!) in cset:1b77ce8ce560. Closing.
#1284 1315215626000000 dread Really? How many times do you want to fix this bug again? For the sake of three lines of test...
#1285 1314269414000000 dread Another instance is when the database tables are not created and you want to both display a sensible error and have an email alert. #1290
#1285 1323178678000000 dread Isn't the point is we don't to raise an exception at all - we want to call a method to send the email and then return from the controller a normal html page. Don't we just need a method to get the error email address from the config and send an email.
#1286 1342436420000000 dread This can happen now, if it hasn't already.
#1289 1317315211000000 dread Discussions have not resolved this either way. Decided to leave it like it is for now.
#1290 1314270894000000 dread Improved on option 2. For requests to Home controller, the first thing it does is check the user. I catch SQL errors that are to do with missing tables and that now produces a template-less error saying the site is off-line and then mentions database initialisation. Done in cset:484e50c4f989 on default for ckan 1.4.4.
#1292 1315999202000000 dread <rgrp> i'm def +1 for removing API stuff entirely from i18n for the present
#1292 1320145677000000 dread Review process of releases and updating strings (pot) and translations (po) files c.f. http://lists.okfn.org/pipermail/ckan-dev/2011-May/000718.html http://wiki.ckan.net/Release_process updated with basic process Get language changing thing to get home page to mostly work - push to key strings translated (or don't advertise the translation?) Welcome message is translated in all languages. Most language have upped their percentage translated significantly and Swedish and Polish added as well. Review long list of strings for improvement Consolidate any? Get rid of i18n for strings in api About 50% of the strings are templates, 30% are in the logic layer and 20% are in the controllers. Removed a handful from ckan/lib/base.py that were only displayed in the api, but the logic layer ones are generally errors that can be displayed in the Web UI as well as the API, so not ideal to remove. Try and remove javascript Tom did this. Add in extensions This could be done. Have raised ticket #1292 for this, next release. #1374 Fix switching to en from other default language Done #1417 Fix browser language detection Done
#1292 1320145732000000 dread These fixes all went into release 1.5.
#1298 1323710593000000 dread Sean says: This is done on the activity stream branch. Just would like David Raznick to review it.
#1303 1314978842000000 dread Adria fixed this in cset:7637d8694388 on default. I transplanted it to release-1.4.3 and have updated thedatahub.org.
#1306 1319797204000000 dread These are expressed as functional API tests - see tests/functional/api/test_action.py. There's not a lot of difference (apart from Paste coming in there) so no real need to duplicate API and logic layer tests, or to switch them to talk directly to the logic layer.
#1309 1315586335000000 dread Duplicate of #1318
#1310 1315820419000000 dread This was fixed cset:e49781cb74fd for 1.4.3.
#1312 1315327743000000 dread The default Content-Type is "application/x-www-form-urlencoded" (that is what curl sends, and what CKAN defaults to). Therefore these characters need url encoding ("=&;" -> "%3D%26%3B"). Here is an example of it working: {{{ (pyenv-ckan)dread@dread-laptop:~/hgroot/ckan$ curl -i http://localhost:5000/api/rest/package -d '{"name": "test7", "title": "Test2 %3D%26%3B"}' -H "Authorization:tester" HTTP/1.0 201 Created Server: PasteWSGIServer/0.5 Python/2.6.5 Date: Tue, 06 Sep 2011 16:36:32 GMT Pragma: no-cache Cache-Control: no-cache Location: http://localhost:5000/api/rest/package/5866a0f5-905a-44ff-91c3-c1489374a73e Content-Type: application/json;charset=utf-8 Content-Length: 571 {"maintainer": null, "maintainer_email": null, "id": "5866a0f5-905a-44ff-91c3-c1489374a73e", "metadata_created": "2011-09-06T14:06:40.735488", "relationships": [], "metadata_modified": "2011-09-06T16:36:32.543461", "author": null, "author_email": null, "state": "active", "version": null, "license_id": null, "resources": [], "tags": [], "groups": [], "name": "test7", "license": null, "notes_rendered": "", "url": null, "notes": null, "title": "Test2 =&;", "ratings_average": null, "extras": {}, "ratings_count": 0, "revision_id": "92ee0b2e-c5e4-47b9-b00c-b8e29662b950"} }}} Alternatively you can set the content type to "application/json" and you don't need to URL encode. e.g.: {{{ curl -i http://localhost:5000/api/rest/package -d '{"name": "test5", "title": "Test &"}' -H "Authorization:tester" -H "Content-Type: application/json;charset=utf-8" }}} To help work out these problems with encoding in the future, I've improved the error message to show the decoded JSON string. cset:55273629a552 which is headed for CKAN 1.5 release.
#1315 1315414218000000 dread Pretty good start added in cset:7a0e7a486df3 on default, aiming for release 1.5. Just needs some loose ends tidying up.
#1315 1320173768000000 dread Tidied up in cset:39daf1afd64a for release 1.5.
#1317 1315416649000000 dread Ok, I think this example may be when searching for "ira". "irina" works me, but "ira" gives: {{{"ira" matched several users}}}
#1318 1315569362000000 dread Problem tracked down to Adria's cset:4f1ca1eb2630 on branch feature-1229-db-out-of-controllers that missed out repoze.who in the redirects after login, when refactoring the user registration. This code went into default on 2011-07-28 for release 1.4.3. Fixed in cset:93731cd3b22e for release 1.4.3 and doesn't affect previous releases. Tests for user registration added too.
#1319 1315904592000000 dread No responses received about this on ckan-discuss, so I'm taking that as no objections to requiring email address on registration. I think that verifying email address is not worth it in this case, because the value of supplying an email address is only for resetting a password. The chance of needing a password reset AND having inputted an incorrect email address is very low, so I don't think it justifies everyone having to do this extra step. If we used email address more then that may change.
#1319 1324036947000000 dread Goes into release 1.5.1
#1321 1319558004000000 dread We should not cache if there are flash messages to show (otherwise they get queued up for the next non-cached page view). Fixed in cset:945e2a6ce74f for release 1.5
#1323 1315594333000000 dread thedatahub has a couple of auth groups with no name. The form allows it! Fixed in cset:bebd0a46ebc6 * form requires the name field * existing auth groups without name can be viewed/edited if necessary (referring by ID) This has probably been a problem since auth groups were added to CKAN.
#1325 1315587113000000 dread Fixed in cset:0dd4d4457c97 for release 1.4.3. Previous Python 2.5 incompatible versions include 1.4.2 - 1.3.4.1 and maybe further.
#1329 1317118323000000 dread I'll take a look to see if I can do any more and then will close it.
#1329 1317404566000000 dread POT looks far better. I couldn't see any outstanding problems. This has gone into ckan 1.5.
#1330 1327060201000000 dread This is simply incorrect. test_authz.py (I assume you mean the one in ckan/tests/functional?) tests that authz works all across the Web UI and API in enough detail. The 'total site lockdown' test - perhaps you mean the TestLockedDownViaRoles class which makes up a small part of test_authz? It was a test added due to an untested use case implemented in IATI, DataGM etc. These are currently active so still need testing. Regarding refactoring the tests to be "proper" then please open a new ticket, providing references to why it is not "proper" to test by searching for text in a web response.
#1331 1325355631000000 dread This appears to be fixed in 1.5, 1.5.1 and 1.5.2a.
#1332 1315999104000000 dread Fixed in cset:860b357a2aec for release 1.4.3 and updated no.ckan.net. This problem was due to {{{[1:authorization group]}}} being translated as {{{[autoriseringsgruppe]}}} and I corrected it to {{{[1:autoriseringsgruppe]}}}. This word has a bold tag, hence the brackets and the number is used to identify which bracketed word is which (when you have more than one). It is a shame that Transifex doesn't draw attention to this problem and Babel doesn't spot the error on compilation.
#1334 1316086680000000 dread cset:b01de680bf86 copes with the problem. Now to find out why revision.packages has some None values - very odd!
#1334 1319798961000000 dread I think this has been a problem since 1.3.3. Fixed in v1.5.
#1335 1315910117000000 dread Done to a basic extent in branch feature-1211-drupal for release 1.4.3. Further tickets: * #1322: Action API improvements * #1315: Document /api/action
#1340 1315991534000000 dread Fixed in default in cset:33856d3ce84e aimed at release 1.5
#1341 1319649573000000 dread The majority of our 6000 users are still spammers. Can we simply elimenate those who've not made any revisions? Even if we take a few genuine users with it, these are pretty dead accounts anyway.
#1341 1319709997000000 dread Reopening - this is still a serious problem. We can't link to the user page because it is so embarrassing: http://thedatahub.org/user
#1343 1316078440000000 dread I don't think email confirmation is necessary - see discussion at #1319
#1346 1316086314000000 dread Fixed in cset:df299e69d456 on default for release 1.5. Problem exists on releases 1.4.3 - 1.3.1 and probably earlier too.
#1346 1316094632000000 dread Also Transposed to 1.4.3.1.
#1347 1316094608000000 dread Fixed on cset:7c3cc0fd6ae4 on branch release-v1.4.3.1.
#1358 1317748211000000 dread You're gonna have massive config files if you do this! Might well be worth it though.
#1360 1317315031000000 dread I believe zephod has removed these from the UI already. I removed them from the search code, tests, docs and provided suitable guidance in cset:5b20ae1673ea which is headed for CKAN 1.5.
#1365 1319815052000000 dread Done on defect-1365-search-params for 1.5. No official releases exist with this broken
#1367 1317394273000000 dread Fixed in cset:d15d81aa63cf for 1.4.3.1.
#1371 1324037905000000 dread Goes into release 1.5.1.
#1373 1319542883000000 dread I think this is due to the etags cache not taking into account whether you are logged in. So the page is cached by squid (which is why it's not seen anywhere apart from thedatahub.org).
#1373 1320174240000000 dread Squid force caching is now turned off on the main page, and etags sorted out in #1388.
#1374 1317895709000000 dread Would be nice also to print the default language first in the list.
#1374 1319648726000000 dread This is fixed in branch defect-1418-i18n for release 1.5
#1378 1324047835000000 dread Went into release 1.5.1
#1380 1324048324000000 dread This has gone into v1.5.1.
#1381 1324037815000000 dread In release 1.5.1.
#1385 1318419159000000 dread Another change needed for this (via Florian): {{{ diff -r 6f27828a4621 ckan/lib/cli.py --- a/ckan/lib/cli.py Wed Oct 12 11:10:32 2011 +0100 +++ b/ckan/lib/cli.py Wed Oct 12 12:23:18 2011 +0100 @@ -132,7 +132,7 @@ pg_cmd += ' -U %(db_user)s' % self.db_details if self.db_details.get('db_pass') not in (None, ''): pg_cmd = 'export PGPASSWORD=%(db_pass)s && ' % self.db_details + pg_cmd - if self.db_details.get('db_host') not in (None, '', 'localhost'): + if self.db_details.get('db_host') not in (None, ''): pg_cmd += ' -h %(db_host)s' % self.db_details if self.db_details.get('db_port') not in (None, ''): pg_cmd += ' -p %(db_port)s' % self.db_details }}}
#1385 1326218703000000 dread This patch has gone in (aimed for CKAN 1.5.2) [master 5b518e5]. Regarding setup to use unix sockets (as opposed to the loopback), I'm not keen, since it requires an extra step in editing the postgres config, for a barely noticeable performance improvement. Thanks for alerting us to it though, Florian.
#1386 1324036930000000 dread These changes go into release v1.5.1.
#1387 1318429228000000 dread I'm generally against validation. When doing any bulk import, it either have to ignore validation errors or lose data. If you ignore the errors, you find you can't edit the dataset without correcting the validation error, and you often end up guessing, just to get your other changes in. All in all, I think it is better to nudge users, with javascript pop-ups in the form, or quality scores, rather than hard-and-fast validation.
#1388 1320174229000000 dread Etags fixed in cset:a3c8d4f7cf91 for release 1.5.
#1393 1318505854000000 dread I changed this: {{{ diff -r 47657581fc30 ckan/tests/__init__.py --- a/ckan/tests/__init__.py Wed Oct 12 17:58:19 2011 +0100 +++ b/ckan/tests/__init__.py Thu Oct 13 12:29:54 2011 +0100 @@ -373,7 +373,7 @@ plugins.load('synchronous_search') def is_search_supported(): - supported_db = "sqlite" not in config.get('sqlalchemy.url') + supported_db = True return supported_db def is_regex_supported(): }}} But there seems to be a problem finding the package when trying to index it: {{{ ====================================================================== ERROR: test suite for <class 'ckan.tests.functional.test_search.TestSearch'> ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/dread/hgroot/pyenv-ckan/lib/python2.6/site-packages/nose/suite.py", line 208, in run self.setUp() File "/home/dread/hgroot/pyenv-ckan/lib/python2.6/site-packages/nose/suite.py", line 291, in setUp self.setupContext(ancestor) File "/home/dread/hgroot/pyenv-ckan/lib/python2.6/site-packages/nose/suite.py", line 314, in setupContext try_run(context, names) File "/home/dread/hgroot/pyenv-ckan/lib/python2.6/site-packages/nose/util.py", line 478, in try_run return func() File "/home/dread/hgroot/ckan/ckan/tests/functional/test_search.py", line 21, in setup_class CreateTestData.create_search_test_data() File "/home/dread/hgroot/ckan/ckan/lib/create_test_data.py", line 66, in create_search_test_data cls.create_arbitrary(search_items) File "/home/dread/hgroot/ckan/ckan/lib/create_test_data.py", line 197, in create_arbitrary model.repo.commit_and_remove() File "/home/dread/hgroot/pyenv-ckan/src/vdm/vdm/sqlalchemy/tools.py", line 110, in commit_and_remove self.commit() File "/home/dread/hgroot/pyenv-ckan/src/vdm/vdm/sqlalchemy/tools.py", line 100, in commit self.session.commit() File "/home/dread/hgroot/pyenv-ckan/lib/python2.6/site-packages/sqlalchemy/orm/scoping.py", line 139, in do return getattr(self.registry(), name)(*args, **kwargs) File "/home/dread/hgroot/pyenv-ckan/lib/python2.6/site-packages/sqlalchemy/orm/session.py", line 614, in commit self.transaction.commit() File "/home/dread/hgroot/pyenv-ckan/lib/python2.6/site-packages/sqlalchemy/orm/session.py", line 385, in commit self._prepare_impl() File "/home/dread/hgroot/pyenv-ckan/lib/python2.6/site-packages/sqlalchemy/orm/session.py", line 361, in _prepare_impl ext.before_commit(self.session) File "/home/dread/hgroot/ckan/ckan/model/extension.py", line 103, in before_commit methodcaller('before_commit', session) File "/home/dread/hgroot/ckan/ckan/model/extension.py", line 38, in notify_observers func(observer) File "/home/dread/hgroot/ckan/ckan/model/modification.py", line 45, in before_commit self.notify(obj, DomainObjectOperation.new) File "/home/dread/hgroot/ckan/ckan/model/modification.py", line 70, in notify observer.notify(entity, operation) File "/home/dread/hgroot/ckan/ckan/lib/search/__init__.py", line 93, in notify package_to_api1(entity, {'model': model}), File "/home/dread/hgroot/ckan/ckan/lib/dictization/model_dictize.py", line 231, in package_to_api1 dictized = package_dictize(pkg, context) File "/home/dread/hgroot/ckan/ckan/lib/dictization/model_dictize.py", line 118, in package_dictize raise NotFound NotFound ---------------------------------------------------------------------- Ran 1 test in 2.734s FAILED (errors=1) }}} Any ideas John?
#1393 1320153590000000 dread Thanks John, reassigning to you for the future.
#1394 1320663304000000 dread I have half a fix for this at the moment.
#1394 1321872676000000 dread Not moved on during fortnight - keen to close this off.
#1394 1323166530000000 dread Not moved yet, but want to finish it...
#1394 1324292900000000 dread Defo for this sprint.
#1394 1325604784000000 dread Done my best here: [release-v1.5.1c c51db1a] Changed "incomplete" to "invalid" but numbering is done in Javascript, so will leave it. I added a test for the main part of the validation (error_dict), but the API doesn't expose the error_summary which the javascript seems to get hold of somehow, so wasn't able to test for that.
#1395 1318525201000000 dread This problem is mentioned here: http://docs.ckan.org/en/latest/test.html but you're right, it really should be solved in the install instructions.
#1395 1320857823000000 dread I added this to the instructions a few days ago to fix this issue for the 1.5 release: {{{ pip install webob==1.0.8 }}} Cheers Sean.
#1396 1330086237000000 dread Went into CKAN 1.6
#1401 1321037805000000 dread Another problem is 401 redirect when your not authorized to do something, like creating a group. The redirect URL is set in the who.ini: {{{ login_form_url= /user/login }}}
#1401 1322066474000000 dread Patch from Augusto Herrmann for the icon URLs: http://pastebin.com/aRBQmftL
#1401 1326460993000000 dread When javascript calls other scripts, it misses the mount-point prefix. e.g. application.js has hard-coded URL /api/2/util/is_slug_valid The URL prefix is stored in request.environ['SCRIPT_NAME'] so perhaps this value should be put in the HTML somewhere for the javascript to get at?
#1401 1326747111000000 dread Links have been fixed on branch defect-1401-mount-non-root. (Instead of using c.site_url the links now just detect the mount url like the rest of the links.) Now it needs the js and css to be fixed too. Reassigning to Tom for this. Brazilians currently waiting for this to be fixed.
#1401 1327665399000000 dread Basic check of the home page shows the CSS has not been fixed: {{{ GET http://127.0.1.1/img/find.png 404 (Not Found) GET http://127.0.1.1/img/share.png 404 (Not Found) GET http://127.0.1.1/img/collaborate.png 404 (Not Found) }}}
#1405 1320152291000000 dread Fixed in cset:e68926926bee for release 1.5. David looked for duplicates on thedatahub.org and found one - http://thedatahub.org/group/lodcloud with http://thedatahub.org/dataset/didactalia. I fixed that by opening the dataset, editing, saving.
#1406 1320930088000000 dread Note, the Atom feeds are still working, with the links still advertised with <link>s home and package pages, but the visual element was removed.
#1409 1324047722000000 dread In 1.5.1 release.
#1413 1322481235000000 dread approximate 'once they log in' as 'whenever they visit their account page' I had imagined that this would be done in the user.py:logged_in method, but doing it in your way sounds like an even better plan. Another thought is: a lot of people might just use the cookie and v. rarely logout/in. Perhaps it could be done when they come back to the site at intervals (e.g. daily)? i18n Yes your strings should be internationalised because they are user visible. You're right that these other strings should be internationalised. I've created a separate ticket for this: #1491
#1413 1323269472000000 dread Code reviewed - looks really good. Do merge in to master if it is not already. A couple of tips: * we all use 'model.User' instead of 'model.user.User' but it's not important for tests... * probably better to use CreateTestData.create_user() to create user objects in the future, avoiding faffing around with creating revisions etc. that are also nastier to debug when they go wrong.
#1413 1330020486000000 dread Went into CKAN 1.6
#1416 1320153507000000 dread This seems to have been fixed in #1229. Added a test for this cset:5d6a3e50fe8f on default to demonstrate it displays a validation error, as rgrp suggests. Note, this ticket is about spammers missing some of the form fields, rather than any normal user usage, where the browser sends all the fields whether you fill them or not.
#1417 1319651617000000 dread Fixed in branch defect-1418-i18n for release 1.5.
#1418 1319539691000000 dread Same as #1374
#1419 1319795920000000 dread This seems to work better today...! myopenid works for me fine with my existing account and by reregistering. My google id works only if I reregister, and I think that is because I hadn't registered before. Closing.
#1421 1319639311000000 dread Fixed in cset:109db2176fac for 1.5 release.
#1422 1319645722000000 dread Done in cset:54fc460602d2 on default for 1.5.1 release.
#1425 1319735012000000 dread Cheers for taking this on Tom
#1425 1326821722000000 dread There is a package_delete logic function to achieve this (and indeed group_delete too). One of the reasons we haven't advertised it with a button is that anyone with 'edit' privileges can go round deleting packages. It's one thing spotting graffiti, but when you delete packages, it takes a lot of effort to spot nefarious deletions. I think should either restrict deleting to sysadmins or have a separate authz-action DELETE-PACKAGE that is not given to the average logged-in user. I believe Wikipedia does something similar - the average Joe can edit, but can't delete pages.
#1425 1330085282000000 dread Went into CKAN 1.6
#1426 1319709236000000 dread Reading http://www.checkupdown.com/status/E409.html I've changed my mind - 409 is ok for an application level validation error.
#1427 1319709925000000 dread Done in cset:ec9a9efc2d40 for release 1.5
#1428 1319804773000000 dread Fixed in a3b0080467d4 for release 1.5
#1430 1324033923000000 dread Fixed in CKAN 1.5.1. Affects all previous CKANs that use SOLR in these circumstances.
#1431 1320084104000000 dread Fixed in cset:711a68a12d90 for release 1.5.
#1433 1324472583000000 dread Has gone into CKAN 1.5.1 release.
#1435 1320930310000000 dread Interesting. What's the reason to justify the effort?
#1435 1323283538000000 dread From IRC today: {{{ <rgrp> dread: btw have a big suggstion -- switch to continuous.io for our buildbot stuff ... <rgrp> openspending have done this and it's a nice setup ... <dread> rgrp: interesting - what's better? <dread> i assume it just starts a script and reports the result? <rgrp> that's right though also has integration with some bakcens (but we probably don't need that) <rgrp> the point is we don't need to boot machines, install and configure buildbot etc (though we may now have automated that ...) <dread> ah, i see, it's in the cloud <dread> did all that months ago, and don't need to setup the machine any more. <dread> if we do it again though, I'm all for it }}}
#1437 1320173496000000 dread This can be fixed with one line of code, so I'm doing it for release 1.5.
#1437 1320173795000000 dread Fixed in cset:5d0bf20a1746 for 1.5. Broken since 1.4.3.
#1440 1320235084000000 dread I've done this. 'db load' now also does 'db upgrade' and 'search-index rebuild'. If you want to debug the load (or don't need the search index, cos it takes ages to do), then you can just do 'db load-only' which does what 'db load' did before. Changed in cset:9e51df5c496b for 1.5 release.
#1441 1320235062000000 dread Fixed in cset:35542f1c60a2 for release 1.5.
← 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 →
Note: See TracReports for help on using and creating reports.