Open ksurya opened 10 years ago
Things to fix:
There may be more bugs, but I'm not founding them yet. I think you have a lot to fix between the two branches either way.
So, a couple of question about the decision you made here:
There is no correct answer for these questions, I just think that you should have an answer for them and understand what are the pros and cons of each solution.
[I will answer 1st comment later on after making some changes] Regarding the 2nd comment:
Two things that I need fix ASAP to continue testing:
If you are not getting error 1, maybe you should try cloning SciPy master, ./quickstart.py, the checkout this branch and migrate (you will have to fake the first comment migration probably... that's something that should be added to quickstart - it could be in the same branch you are fixing the search problem).
Ok, got voting to work. Number 2 still holds, though. Other things to improve:
That's a lot to do for now.
[2] The JS code waits for server to complete request. So, we see small lag which sometimes can be observed clearly. This kind of method ensures
We can also do it in the other way First update the data from the client code itself. Then, edit them only if they are different from server.
Please let me know if you have any better idea.
As I said, I think it would be better to do it the other way. As it's not important data, it's not really awful if we are showing incorrect data to the user.
On 13 September 2013 18:16, Surya Kasturi notifications@github.com wrote:
[2] The JS code waits for server to complete request. So, we see small lag which sometimes can be observed clearly. This kind of method ensures
- To display only when its successful
- The total number of votes is refreshed. If some other user votes in the middle, its also counted.
We can also do it in the other way First update the data from the client code itself. Then, edit them only if they are different from server.
— Reply to this email directly or view it on GitHubhttps://github.com/scipy/SciPyCentral/pull/134#issuecomment-24409908 .
[1] I have observed it but didn't know the reason. But after making other changes, I didn't see it
[3] Roughly, its taking 250-400ms for a vote. Do you think its huge?. I tried to reduce number of queryset iterations required while counting number of up votes, down-votes. I think this should help a bit but not too much.
[4] I removed the arrows. Its clear that way
[5] It does not make sense voting their own submissions. Besides, if a person has 50 submissions, and votes on each of them, it gives him 250 reputation on his profile!! It seems to be considerable abuse
[7] Right now, we only see latest revision on home page while search queries contain all revisions!
[8] I have actually posted about it in mailinglist when I started thumbs but it didn't get any considerable discussion. I think if we are going to create revisions even for a typo in existing submission, carrying reputation makes sense. Otherwise, it may not be necessary.
[12][13] They are ordered based on wilson score. If you have changed "reputation" field from admin, try changing "score" fields and see. It has to work. Technically, the score fields range from 0-1
[15] This data is created based on "pagehits" app and its data. Sometimes, if we make less hits on our db, we don't see any submissions there. However, if there are hits and we delete respective revision from db, the homepage raises error. This has to be fixed though.
I'm not even sure what you call score, and what you call reputation. You use both terms interchangeably in comments and code, and you even use wilson score
some times, which makes everything more confusing. Choose the terms you want to use for each, and do so consistently. For example, reputation can be the simple sum of upvotes and downvotes, and score the wilson score.
I'm not sure if I already said it, but have you tried creating indexes for Thumbs to see if we can improve the performance?
Ok, I think that's it for now.
:+1:
Maybe something can be borrowed from: https://faq.i3wm.org/questions/
Reputation system for the site