Axtell

Downgoat

00:17

can someone else approve this: github.com/Mego/PPCG-v2/pull/19

Anonymous

@Downgoat Ffs stop

Downgoat

If I have like:

a href="/user/logout?redirect={{ request.url }}">

is there a way to escape request.url so the user can't do like " onclick="evil js" as part of their URL

Anonymous

@Downgoat Append |e

Downgoat

TIL html escape work in attribute

Anonymous

00:33

Yep, you can pipe any string through e. It just does normal HTML escaping.

Downgoat

00:47

WIll we have rep?

Anonymous

In what manner?

Downgoat

just like points that people gain when an answer is upvoted etc.

Anonymous

Yes

Anonymous

Haven't decided on the scale, yet, though

Downgoat

oh ffs I just did a stupid

ok screw it I'll just merge the whole thing

Anonymous

00:56

You worry me when you say things like that

Anonymous

You should set up your IDE to auto-commit every 5 minutes or something

Pavel

Is there a way to git stash data and then git stash pop it to multiple branches?

Downgoat

how would I use url_for for resources which don't have routes. do you mean like url_for('templates', filename='svg/user.html')

Pavel

So like git stash && git checkout goat-backup && git stash pop && git checkout <working-branch> && git stash pop

Downgoat

@Pavel store a diff and apply the diff to multiple branches

git add -A && git diff @ > stash.diff && git checkout b1 && git apply stash.diff I think

or you might have to do git reset instead of git add -A not sure

Pavel

01:01

well git reset will just nuke the changes you're backing up so that seems wrong.

Downgoat

no it unstages things

Anonymous

@Downgoat You could just add a route for the base dir and then pass the filename as a GET param

Pavel

git reset reverts your working directory to the last commit

Downgoat

@Pavel you're thinking of git checkout/reset --hard

@Mego that would give a URL for like <a> tag, would it work for a flask compile-time {% include %}?

Anonymous

@Downgoat It should. Routes are loaded before templates are initialized.

Downgoat

01:09

url_for('/', filename='includes/header/authmanager.html')

that seems to give an error that there is no endpoint /

oic

oh wtf SE uses old open id protocol ;_;

Pavel

It's like a decade old by now

Downgoat

ok well time to add oauth then I guess :|

oauth doesn't provided any identity though

do we know if SE uses v1 or v2?

Anonymous

@Downgoat It should be something like url_for('headers', filename='authmanager.html'), and then have @app.route('/headers/<filename>') def headers(filename): return whatever

Anonymous

@Downgoat v2. They have documentation here

Christopher 2EZ 4RTZ

why are you people so smart :()

Pavel

01:18

@Christopher2EZ4RTZ Been doing this for a long time

Downgoat

@Christopher2EZ4RTZ you're saying this to the ~~guy~~ goat who just did sudo rm -r static/js

Pavel

Experience unfortunatly does not bring with it carefulness.

Christopher 2EZ 4RTZ

@Downgoat WHY DO WE LET GOATS HAVE LINUX??

Downgoat

Would we store the refresh token in the DB?

that wouldn't work with logging in again though

would that mean we have to implement seperate oauth route for every. single. provider. we want to support?

Anonymous

@Christopher2EZ4RTZ If you don't have anything constructive to contribute, then don't.

Christopher 2EZ 4RTZ

01:20

Do you have anything simple that I can do? @Mego

Anonymous

Here is an example implementation of SE OAuth2's flow in Python @Downgoat

Anonymous

@Christopher2EZ4RTZ Look at the project board on GitHub. There's a few things left for basic API stuff.

Downgoat

@Christopher2EZ4RTZ yes, unittests

Christopher 2EZ 4RTZ

@Downgoat I have read many reddit memes about how hard unit tests are. Idk if they are true but I am going with mego

Pavel

I'm currently taking an extra semester of math outside of regular school. I procrastinated on it and had to compress like two months of curiculum into the past week, but I'm getting it completely finished today so I'll be able to work more from now on.

@Christopher2EZ4RTZ They're really not hard, unless you're using Junit. Then it's a PITA.

Anonymous

01:23

@Christopher2EZ4RTZ Unit tests are easy. Being sure that you're done writing unit tests is hard. Automatic code coverage tools like coveralls helps with that.

Downgoat

@Mego It seems like you're using an access_token as a user's id in the DB but that will change when the user logs in again, no?

Anonymous

@Downgoat The access_token bit from the POST response is the part we're interested in. If we set the no_expiry field, we don't ever need to change it, so logging in won't require redoing the auth request.

Anonymous

@Christopher2EZ4RTZ This is straying really far from this room's topic.

Christopher 2EZ 4RTZ

@Mego I came from TNB so i fix it back

Anonymous

→ 7 messages moved to Trash

Pavel

01:29

Not exactly on topic for TNB either but I don't think anyone cares about what's on topic there.

Christopher 2EZ 4RTZ

That was the joke :|

Downgoat

@Mego how would we know which user was just authorized?

would the same access_token be returned

Anonymous

@Downgoat Authorization would be done in a single session, so then we can tie an access token to a user_id (our internal IDs)

Downgoat

btw I just updated the PR to address your comments

what's the difference between a post and an answer/comment

Anonymous

01:51

@Downgoat In SE terms: post = question, answer = answer, comment = comment

Pavel

It's just lovely because SE refers to questions and answers collectively as posts.

Anonymous

Yeah, I wanted to separate them out for a better hierarchical DB setup

Pavel

Given we have a site dedicated to code-golf, I propose we call them challenges and solutions.

Downgoat

oh ok

@Pavel we can call it whatever on frontend but I think on backend it would be better for it to be more generic than less = better

Anonymous

@Pavel a) Competitive programming in general, not just code golf. b) I want this to support Q&A related to competitive programmming, too.

Pavel

01:54

@Mego You mean like tips? That should be it's own thing entirely.

Maybe like wikis for langs

Downgoat

do we want to do an infinite scroll or pagination?

Anonymous

For more generic, we could rename answer to response, but then that will conflict with the HTTP response stuff

Anonymous

@Pavel I'm thinking we should have 3 parts to the site: main for challenges, meta for meta, and q&a for q&a

Anonymous

@Downgoat Probably pagination, to be nice to mobile users

Pavel

We could also detect if the user has a mobile browser or not.

Downgoat

01:57

responsive > mobile detection

idk how to do paginated SQL query

Pavel

And infinite scroll is nicer for desktop users

Downgoat

Since a post list is discovery, infinite scroll would be better from UX perspective

Anonymous

But with infinite scroll, it's really hard to get to the last result. Imagine if the Sandbox had infinite scroll, and you actually had a reason to look at the last page.

Downgoat

btw what do we want on home page

we could go the way of quora/SE with a hot challenge feed

Pavel

@Mego Just press the end key

Anonymous

01:59

@Downgoat Giant penguin. When we think of something better (unlikely), then we can change it.

Pavel

@Mego See this is why Goat is doing frontend.

Anonymous

@Pavel Which either scrolls you down to the end of the current "page", or scrolls you to the end of the content and surprise you have potentially a few hundred things loaded. Also where is the end key on mobile? :P

Downgoat

bueno?

Pavel

@Mego There isn't one, which is why I suggested detecting if it's mobile or not and doing infinite scroll on Desktop and pages on mobile.

@Downgoat I expected it to look like utter crap but that actually looks nice wtf

Anonymous

@Downgoat Perfect.

Anonymous

02:01

@Pavel A symmetric UX between desktop and mobile would be better.

Downgoat

^

Pavel

Idea: Do pages, but you can set the number of answers per page, and the default is higher on desktop so that for most challenges you'll be able to see all the answers on one page.

Anonymous

Maybe I'm just old fashioned, but I feel that infinite scroll is inferior because it takes control away from the user. Imagine a user going "oh I remember that, it should be on page 87/90". With pagination, they go directly to page 87. With infinite scroll, they give up around page 5.

Anonymous

@Pavel Idea for building off of that: posts are collapsable to just the header, and they start collapsed by default.

Downgoat

eehh

I was thinking things like explanation are collapsed by default

but like header + code is shown

Pavel

02:05

@Mego Since most posts would be one line anyway, it could also just collapse the explanation.

Downgoat

code is limited to like 8 lines maybe unless "see more" is click

Pavel

Ninja'd

Anonymous

That could work, too

Anonymous

Collapse the code inside of the code box, and the explanation. More information in less space.

Downgoat

idea: have like two buttons, one for like typical se-esque list view of answers, another views all answers in like grid so you can see like 12 at a time

Pavel

02:06

Anyway, you know how in SQL we can make a query like SELECT * FROM answers WHERE post_id = <post_id> LIMIT 10 to get the first 10 entries? How do you get the next 10 entries, without doing LIMIT 20 and manually discarding the first 10?

Anonymous

@Pavel You would do the request normally, without the LIMIT 10, but only scroll through the rows 10 at a time on the server side

Downgoat

@Pavel .yield_per(10) in sqlalchemy

Pavel

Got it.

Downgoat

Do we want a side bar or something

Anonymous

Probably eventually, with featured/hot posts from each section

Downgoat

02:29

where should the "Post Challenge" button be

should it be in sidebar, or in header, or somewhere else

Downgoat

02:43

@Mego fixed to just POST

Anonymous

04:19

@Downgoat Approved and pulled

Anonymous

Apparently GitHub can track dependencies now, unless they're in requirements.txt

Downgoat

oh that's cool

:| more JS than python apparently

btw in the new repo you need a GOOGLE_CLIENT_ID (from google dev portal) to use for google login support

Anonymous

I'm guessing that's something we'll need to load into real config.py instances

Downgoat

@server.route("/user/<int:user_id>", methods=['GET'])
def get_profile(user_id):
    try:
        uid = int(user_id)
    except:
        return

    return user.get_profile(uid)

is the int() part necessary?

Anonymous

04:51

@Downgoat It shouldn't be

Downgoat

Anonymous

05:27

@Downgoat 8/10 no giant penguin

Downgoat

05:42

Idea: 'parent tags'. e.g. fibonacci would have parent tag sequence and so all fibonacci challenges would go under sequence also

Anonymous

@Downgoat That's a good, complicated idea

Downgoat

btw what do we want to do for the editor we can:

1) use plain textbox
2) write our own markdown editor
3) use [Quill](https://quilljs.com)
4) use another existing markdown editor

Anonymous

I've used Quill before. It's nice.

Downgoat

Downside is I don’t think it does markdown

Anonymous

Doesn't look like it

Downgoat