currently all the social logins are broken.

also it would be good to be able to reconcile any of these logins to the same account, through verification or some such.

if comments can be sorted via multiple means (e.g., votes, similiarity), that might provide a way to merge or consolidate comments into better answers.

Neither /login nor /browserid/login routes seem to work when commenting and logged out.

Also, clicking on ¨New post¨ when not logging in does not find the /login route.

• search/coordinate search across biostar sites
• api access for post/search aka stackoverflow
• RDF export
• ORCID integration
• Weighted metrics
• Prompt for topic when posting
• Tag filtering
• Write a good set of deployment recipes
• Mobile-friendly site

settings.py to be unversioned and present in .gitignore and settings.py.sample for general use.

Eases deployment.

When subscribing to the RSS feed (http://neurostars.org/feeds/latest/) I noticed that all the links point back to the base URL of example.com.

Looks like this was brought up before: ialbert#80

atm would lead to Bad Request (400)

The pill bar is shown in case of anonymous user
Links in this bar lead to pages that generate an Internal Error:
http://neurostars.org/show/messages/
http://neurostars.org/show/myvotes/
http://neurostars.org/show/mybookmarks/
http://neurostars.org/show/myposts/

Find attached a screenshot with the pill bar highlighted in red

Right now trying to post comments in neurostars results in the following (example) 404:

http://neurostars.org/new/comment/f2qIJx936eyyzGhioTajkxmlx10hGeGb/

Additionally, clicking repeatedly on "comment" generates several div=edit-action sections.

I suspect that Django's CSRF is getting in the way:

https://docs.djangoproject.com/en/1.6/ref/contrib/csrf/

Although it is correctly enabled in MIDDLEWARE_CLASSES in settings.py via 'django.middleware.csrf.CsrfViewMiddleware',...

@ialbert, sorry to bother you again, but if you have some insight in this matter, it would be very appreciated :)

Figure out how the Django templating system in BioStar interacts with Mozilla persona. Just placing the django_browserid placeholders in nav.bar.html did not work properly.

As commented by Giovanni in: https://www.biostars.org/p/99956/#101306

The top categories header needs a better placement and customizability (other than editing some .py files)

When I try run the project following the instructions, I get this error:

"django.db.utils.OperationalError: no such table: django_site"
I have the correct default.env, and this is my "pip freeze":
Django==1.6.2
South==0.8.4
amqp==1.4.4
anyjson==0.3.3
billiard==3.3.0.16
celery==3.1.9
django-celery==3.1.9
django-debug-toolbar==1.0.1
django-markwhat==1.1
django-pipeline==1.3.22
futures==2.1.6
kombu==3.0.13
pytz==2013.9
sqlparse==0.1.11

The complete traceback here: https://gist.github.com/hannicolas/9460328

I haven't heard anything from neurostars RSS for a while even though my r2e checks every 10 minutes. and few minutes ago all of a sudden got like 30 of those announcements, probably dating all the way back. was there a restart or smth

I received them around Mon, 13 Jan 2014 05:20:02 -0000

• Start by integrating ORCID as an OAuth provider in django-allauth.
• Once we have that, I guess the first step would be to pull some data from ORCID and show it on the users’s profile, we can discuss cooler use cases in GitHub issues later, but here are some hints on what’s possible:

http://support.orcid.org/knowledgebase/topics/19247-rest-api

Ensure that the site looks and works smoothly on mobile phones. Preferably using Django plugins for that:

https://pypi.python.org/pypi/django-mobile

When running a local copy of the site from the develop branch, sqlite3 operations fail due to the new autocommit policy in Django:

https://docs.djangoproject.com/en/1.6/topics/db/transactions/

Does not happen in production since there postgresql is in use.

Your database backend doesn't behave properly when autocommit is off. Turn it on before using 'atomic'.

The post gets created anyhow...

The current design is a bit cluttered with tags, features and latest user(s) activity. Study other ways to present the information, following today's design practices.

One good example of such simplicity would be Google Scholar and perhaps Stackoverflow itself, we have to find a good balance between simplicity and usability.

As neurostars.org admin, I'm regularly getting the following email notifications:

[Django] ERROR: Invalid HTTP_HOST header: 'www.daydaydata.com'.
You may need to add u'www.daydaydata.com' to ALLOWED_HOSTS.

No stack trace available

Request repr() unavailable.

We should filter those out as explained here, adding it to our deployment process on ansible-biostar:

http://stackoverflow.com/questions/18220519/how-to-disable-djangos-invalid-http-host-error

A standard statement about INCF should be placed under about.

Missing requirements in neurostarts.org branch.

django-openid-auth
django_compressor
python-openid

two issues:

1. login should describe what mozilla persona is - people won't click the link to find out and get confused.
2. there should be a direct link to github issues on the neurostar banner so people can report bugs easily/directly.

@nimiq, apparently there was a bad merge with the about/faq pages on neurostars. The production page shows generic info from biostars, not neurostars:

http://neurostars.org/info/about/

The following html templates had the correct information. Maybe it's worth to point/refactor them to "themes" folder to avoid future accidental merges.

https://github.com/brainstorm/biostar-central/blob/neurostars.org/main/server/templates/pages/faq.html

https://github.com/brainstorm/biostar-central/blob/neurostars.org/main/server/templates/pages/about.html

During the OHBM hackathon in Berlin, a few users expressed their interest in having a more interactive version of the FreeSurfer mailing list:

http://www.mail-archive.com/[email protected]/info.html

That would allow users to:

1. Quickly search through archives.
2. Rank answers for relevance (collaborative filtering).
3. Filter by tag(s).

For that import to happen we should:

• Ask the original archive owners if that's ok for them.
• Identify the attributes we wish to migrate (usernames, thread structure...).
• Pre-filter relevant questions to have a good bootstrapping.
• Setup an email mechanism similar to Gmane so that new emails get into neurostars automatically, not requiring periodic imports.

@satra, @nicholsn, feedback is welcome. I thought about starting with this:

http://stackoverflow.com/questions/22006616/how-convert-an-mbox-to-a-json-structure

While @nimiq is finishing up the deployment roles:

https://github.com/nimiq/ansible-neurostars.org

I'm experiencing this last strange issue, just about ready for neurostars.org beta launch ;)

http://stackoverflow.com/questions/18005574/django-1-5-get-404-on-static-files

@satra, @richstoner, @ialbert.

@brainstorm, @satra, @ialbert

I've been working on the new deployment strategy based on Docker and Vagrant last week, but I didn't get any good result and I have some concerns about the entire strategy.

In brief

Building a development environment for Biostar is currently very easy: create a Python virtual environment, install the requirements, change the settings in order to use PostgreSQL (if you want to, otherwise you can keep using SQLite), load the environment variables and run some Django management commands (via the biostar.sh script).

Deploying Biostar to live is more complicated and requires to setup Nginx, Gunicorn, PostgreSQL, Redis, Celery, Supervisord and Elasticsearch.

The reason why we don't need to install all these tools in our development environment is that we don't care what is the tool used for caching or full text searching f.i. as long as it behaves like a cache or full text search engine. This principle lets us keep our development environment very simple and light so we focus only on coding. In the rare situations when we have to test those special features we use a staging website (which is an exact copy of the live environment).

Now we want to build a new deployment strategy which can build a development env and deploy to live in the same way, using Docker containers. Docker is a really cool technology and I am very excited to learn such things, but I found so many complications that I am wondering if I am on the right track. I'll dig into more details later on, but the general idea is: a lot of code is required to configure such a deployment strategy, even some workarounds are required because this technology is still not perfect, and I am worried about the maintainability of the the entire process.

More details

• Docker works only with linux OS, so if your development machine is a Mac OS, you need to add boot2docker and Vagrant to your toolchain. This adds some complexity to your environment and makes it heavier so it must be worth.
• Docker containers are meant to contain one or a few tools, so in our case we need to use at least 4 containers. This requires us to write a lot of code to configure Docker and Vagrant. Take a look at this Vagrant file which defines only 2 simple containers (one for PostgreSQL and one for Django). I'm afraid that all this code will easily become unmaintainable.
• Docker containers must be linked together: f.i. the container with the code and the webserver needs to know the IP address and the port of the PostgreSQL container, the Elasticsearch container and maybe the Celery+Redis container. These links add complexity to the Vagrant configuration file and also the Django settings file because we need to use cryptic environment variables like:
  get_env("PG_PORT_5432_TCP_ADDR")
• Forwarded ports are hell. You want to forward some ports from a container to your dev machine, f.i. the port 80 (or 8080 or 8000) of the webserver to see the website in your browser or the PosgreSQL port. Exposing a port from a Docker container is easy, but then (assumed that you are on a non-linux machine) you have to expose it again from Vagrant to the host machine and this doesn't work currently, because Vagrant uses a boot2docker light VM which has an issue. Of course there is a workaround for this, but again we are adding complexity.
• Containers are sort of lightweight wrappers around an app, so they have no SSH access. If you want SSH access to a container you have to add code to install the SSH server and distribute private keys from Vagrant to the Docker container. And you need SSH access at least to manage Django (create migrations f.i. or any other management command).
• With no SSH access to a container there is no way to view logs or configuration files. You might want to see the PostgreSQL or Celery log files, but you cannot if the container has no SSH access. So you will end up configuring the SSH access for all containers.
• Usually passwords are managed via environment vars loaded with a source command. With the new strategy we need first to share these environment vars with Vagrant and then with the containers: I don't know what is the best way to do it yet but I have some concern about security.
• Performance: the code is physically stored in our dev machine. We share it with Vagrant which shares it with the Docker container where the webserver is. This introduce a lag. I did a few tests with my Macbook Pro (with 4 GB RAM, bought in 2010) and the lag was annoying. Vagrant with only 2 containers for PostgreSQL and Django takes at least 0.5 GM RAM.
• Docker is a great technology and it is getting popular very fast, but most of the info in the Web describes very basic configurations with 1 or a few Docker containers. It is hard to find info about more complex systems.

Conclusion

Docker and Vagrant are cool technologies but there are many complications involved. I am not afraid of complications, I can manage this, but I am afraid of 2 things: first this will take some time, maybe weeks and this means less time to developing new code and features for Neurostars; second the result might be not easy to maintain.

What do you think? Fortunately we can discuss it live in Berlin in a few days!

Write a .travis.yml file that runs a testsuite:

https://github.com/ialbert/biostar-central/tree/master/biostar/server/tests

Possibly selenium too if time allows us to do so.

Setup one of those badges, preferably lower right, so that it does not affect the overall styling.

http://www.incf.org/images/ribbons/ribbon-upper-left/view
http://www.incf.org/images/ribbons/ribbon-upper-right/view
http://www.incf.org/images/ribbons/ribbon-lower-left/view
http://www.incf.org/images/ribbons/ribbon-lower-right/view

Engagement with neuroscience community

This issue was created by @shirishgoyal via Mozilla Science Lab Collaborate

Deployment of the web site should be smooth, and maybe pythonic :)

I have a set of ansible roles that need improvement, but that basically deploy a whole neurostars from scratch.

This error occurred because django.contrib.markup is deprecated in the new version of Django ( Django > 1.5 ). https://docs.djangoproject.com/en/1.5/ref/contrib/markup/

to_xml method is does no longer exist in later html5lib api.

• Integrate openid/mozilla persona login with roles in django
• provide moderating/editing permissions to editors

Have the whole contents and scientific data exposed as a git-annex endpoint(s), so that users can fork and remix the contents easily.

http://git-annex.branchable.com/

In Django 1.6, one should create a custom user model. Fix that.

@ialbert, you might be interested in knowing this bit too for BioStars revamp...

Currently there's arguably many users that just miss the "getting notified by email" drop down menu on the questions.

Exposing it on a different way so that is more obvious to the user would improve the adoption of this feature (which helps on keeping users engaged on the site).

Don't exists any reference to browser_id package in settings.py.sample . but the package(templatetags) is being used in template/base.html. http://d.pr/i/kyuK

browser_id: https://github.com/mozilla/django-browserid/blob/master/docs/setup.rst

checking..

from @satra:

Much like google generates a related items box based on their knowledge graph, i think the best use of semantic technology would be to create a related box for any question.

This box may contain pointers to code, to websites, to papers, to other relevant discussions.

So this would entail having a box next to a post and building up a database of related areas (like freebase and using freebase where relevant).

similar to biostars, create a planet feed from relevant neuroscience and other rss sources for neurostars