mrgambal / vulyk Goto Github PK

Please sign here https://travis-ci.org/mrgambal/vulyk and authorize Travis to get push hooks, so project will be rebuilt on every push.

Thanks!

We need "default" group which is always available and assigned to all newly created users.

Groups managing with CLI is a pain now, and 80% of use-cases will be “just allow users to do some work”. Having default group will simplify that to one command: manage.py group addtype ...

Each plugin might have optional JS and CSS assets that it should expose to the framework

We have webassets/Flask-webassets on board already to work with merges that belongs to the framework (check assets.py for details).

On plugin pages we want to include both types of merges, those from framework and those from current plugin (so plugin might talk to frameworks methods defined in Vulyk object).

Please investigate if we can use webassets to manage plugin's assets as well. Take into account, that at least CSS assets for plugin aren't mandatory.

Please reflect that in documentation.

Silly bug... 😞

We need to migrate concept of batches from unshred.it.

Idea here is following: Tasks are imported in batches. Tasks in the feed for the user is also ordered by batch, so user will receive all tasks from batch that comes first before next batch will be started.

This is a requirement because we don't want our results to be spread across 20000 of declarations with average of 0.5 response for each declaration. We want first 200 declarations (say, from ministry of ecology) to be done first with 3 responses per decl.

We need a simple script that will be installed alongside with package and can create an empty project with file structure like that:
Procfile
app.py
local_settings.py
----probably (optionally?)---
heroku related stuff.
uwsgi related stuff.
.hgignore
.gitignore
plugins/demo_plugin.py

It should be described in docs as well.

Basically we need super simple analog of django-admin.py or scrapy.py so newcomers can easily start playing with the system.

I am trying to install and run. I get an error executing this installation command:

pip install -e ../vulyk-declaration

I am running Ubuntu on WSL2 Windows 10

Description: Ubuntu 20.04.1 LTS
Release: 20.04

Building wheels for collected packages: rjsmin
Building wheel for rjsmin (setup.py) ... error
ERROR: Command errored out with exit status 1:
command: /mnt/c/Projects/sandbox/venv/bin/python -u -c 'import sys, setuptools, tokenize; sys.argv[0] = '"'"'/tmp/pip-install-e315onj0/rjsmin/setup.py'"'"'; file='"'"'/tmp/pip-install-e315onj0/rjsmin/setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(file);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, file, '"'"'exec'"'"'))' bdist_wheel -d /tmp/pip-wheel-cxcv1bo0
cwd: /tmp/pip-install-e315onj0/rjsmin/
Complete output (12 lines):
running bdist_wheel
running build
running build_py
creating build
creating build/lib.linux-x86_64-3.8
copying ./rjsmin.py -> build/lib.linux-x86_64-3.8
running build_ext
building '_rjsmin' extension
creating build/temp.linux-x86_64-3.8
gcc -pthread -B /home/dandrade/anacondawsl/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -DEXT_MODULE=_rjsmin -UEXT_PACKAGE -I_setup/include -I/mnt/c/Projects/sandbox/venv/include -I/home/dandrade/anacondawsl/include/python3.8 -c rjsmin.c -o build/temp.linux-x86_64-3.8/rjsmin.o
unable to execute 'gcc': No such file or directory
error: command 'gcc' failed with exit status 1

ERROR: Failed building wheel for rjsmin
Running setup.py clean for rjsmin
Failed to build rjsmin
Installing collected packages: rjsmin, translitua, wheel, unicodecsv, XlsxWriter, vulyk-declaration
Attempting uninstall: rjsmin
Found existing installation: rjsmin 1.1.0
Uninstalling rjsmin-1.1.0:

Environment:

https://vulyk.herokuapp.com, any browser

Steps to reproduce:

• Authorize via Facebook
• Load any declaration
• Scroll to the bottom, Розмір доходу в іноземній валюті

Actual result:

Iнший тип валюти is either enabled or disabled but visible at all times

Expected result:

Only show it when a Інша option is previously selected

Environment:

https://vulyk.herokuapp.com, any browser

Steps to reproduce:

• Authorize via Facebook
• Load first declaration
• Click on the PDF link

Actual result:

In some states the link color will change to #fff and it will be invisible on the #fff background, which is slightly confusing.

Expected result:

Do not change the typical browser-based link coloring behavior

we have 2 plugins requirements in requrements.txt

-e git+https://github.com/hotsyk/vulyk-declaration.git@master#egg=vulyk_declaration
-e git+https://github.com/hotsyk/vulyk-tagging.git@master#egg=vulyk_tagging

At first sight but when I try to dry run setup.py develop I get something like:

Processing dependencies for vulyk==0.1.0
Searching for vulyk-declaration
Reading https://pypi.python.org/simple/vulyk-declaration/
Couldn't find index page for 'vulyk-declaration' (maybe misspelled?)
Scanning index of all packages (this may take a while)
Reading https://pypi.python.org/simple/
No local packages or download links found for vulyk-declaration
error: Could not find suitable distribution for Requirement.parse('vulyk-declaration')

I found the reason: some say that pip runs setuptools, which calls http://nullege.com/codes/search/setuptools.package_index.safe_name so I want you to just rename your plugin-repos (it's safe now) and fix existing links.

In case if you have better solution - you're always welcome.

If user has only one task type available, we should redirect him from task_homes straight to this task

Not particularly a problem but user should get some confirmation that their action was successful (a notification or a redirect to a page).

Please define both, a structure of plugin directory (including optional assets) and min/max set of classes that user must/may define to write his own plugin and methods of those classes as well (check for those that marked as NotImplemented in abstract classes).

Reflect that in docs.

We need to display user's progress (amount of tasks of this type done by him since reg, since beginning of the session, his position in rating for this task) somewhere in the footer. Ideally we'll need a scoreboard as well.

I've added a new branch dc-not-so-dummy-task that contains file with tasks, which is importable.

I've created frontend part (it might be slightly changed but essentials are there), it's available here:
https://dl.dropboxusercontent.com/u/4366002/pos_demo.html. As you might see in the code, it's already can render tasks using handlebars.

Once #9 and #10 is in, it would be nice to integrate it (including required assets).

We can export only whole amount of reports related to certain tasks type, but would be great to have an ability to diversify this process and break it down to batches.

Right now ties are broken randomly, instead we want to have places in rating to be allocated in this way:
if two users has same amount of reported answers, they'll share the place.

In addition to current place and the number of tasks done we want to show the number of tasks that are left. It's a bit complicated, as we have batches, skipped tasks and concurrency.

Some good ideas needed.

Right now plugin only has a system name and help text.

We need also human readable name, logo (will require changes to static files scheme from #23) and short description. Logo and short description should be optional.

Timestamp for Report.createdAt is statically calculated at the application start and doesn't update in future.
We need to move it out from default-parameter to the place where reports are being created.

At the moment the only way to do that is to check attached task for task_type, this doesn't seems right to me.

Environment:

https://vulyk.herokuapp.com, any browser

Steps to reproduce:

• Authorize via Facebook
• Load first declaration

Actual result:

404 Missing favicon: https://vulyk.herokuapp.com/favicon.ico

Expected result:

Some icon added, at least a stub one

Please check and enable python 3 as the runner for documentation builds here on RTD.
The project is targeted only for python 3.4+, however its documentation is still being built using

python2.7 -mvirtualenv --no-site-packages --no-download /home/docs/checkouts/readthedocs.org/user_builds/vulyk/envs/latest

which causes:

  File "/home/docs/checkouts/readthedocs.org/user_builds/vulyk/checkouts/latest/vulyk/models/task_types.py", line 174
    yield from map(lambda a: a.as_dict(),
             ^
SyntaxError: invalid syntax

and lacunas in the final documentation.

Current version of docs is definitely as old as mammoths 💩 , and doesn't reflect any aspect of the application at the moment.
We must update it due to need to easily get newcomers involved into the workflow.

This should display nice table, sorted by batch name.

For each batch it'd be nice to have a number of closed tasks, % of completion (total number of answers is counted as number of tasks in batch times redundancy setting). Each completed task is counted as redundancy number of answers (so we won't have problems with tasks that has more answers than redundancy or tasks that was closed because of inner logic of plugin before reaching redundancy).

Breakdown by number of completed tasks would be also nice!

We need to at least exclude generated js/css from here

At the moment 2/3 of tasks we have already processed. So in few days we'll ran out of work.

I think here we need to cover two cases:

• When static is served through flask app. To make it work we'll need to wrap our plugins into _PackageBoundObject, and expose their send_static_file in urls config just like flask itself does for blueprints. An alternative here is just to update urls config without wrapping.
• When static is collected during deploy with extension like Flask-Collect. Files collected this way will be physically placed into given static folder and can be served with nginx directly (and this will save some CPU and provide better caching etc). I'm not sure if Flask-Collect plays well with Flask-Assets, but if we are going to cover that case too, we will need to create customer Flask Collect storage (https://github.com/klen/Flask-Collect/tree/develop/flask_collect/storage) that will look not only into flask app and it's blueprints but also into plugins.

Third way is to refactor plugins to be a proper blueprints but I'm not sure I have enough courage to do that now.

I'll play with no. 1 now.