klenin / cats-main Goto Github PK

This is an umbrella issue for moving using AJAX requests in various parts of UI.

Since they depend on testsets

User expierence improvance regarding whole system

• Move everything to AJAX isn't bad idea, jquery-pjax is probably best way
• We shouldn't display any system information, an example "Access Prohibited", "Incorrect.." blah blah, as like responses with error code 500, that definitely helps attacker detect system breaches
• Is json option in some places realy needed? If it realy needed, we should change src keys, so they will not equal keys in database, and return same information, that could be fetched from page if it was html
• As for now current contest doesn't saves(if you logout and them log-in you will be selected Archive contest)(isn't it?), why doesn't save it?

User expierence improvance for teams page

• It was really hard for me to locate pagination, intuitively it should be in down-right corner(why not just put a copy of it?)

User expierence improvance for console

• Currently, it displays all runs for same task, why just not display last from that, or if needed make something like "spoiler"
• Auto-opening new run is useless, since modern browsers will block popup windows if them wasn't open with one of functions that originate from user event(click etc.), we can exclude domain from that behavior, but it's makes no sense

Change 'lock_counter' into 'accept_mode' tristate with values 'any', 'pinned', 'none'.
'pinned' mode should accept only requests pinned to this judge.

See $cats::penalty, contests.penalty

Use links to modern checkers instead.
Maybe mass-modify old problems.
select count(*) from problems where std_checker is not null
196

We need to set http.proxy for git to access external repositories.
I have added the proxy variable to Config.pm.template.
Unfortunately, git config --global invocation fails at least on Debian due to the fact
that www-data user does not have a HOME set.
Options are:

1. Figure out how to config Apache for setting HOME
2. Use git config --system in installation script
3. Use something like http_proxy=$proxy; git ... for git invocation

The goal is to display graphs of memory/time depending on input size.
Since problem packages do not currently contain explicit input size data,
the method used should work even in the absence of such data.

"Resources" in question are user time, wallclock time, peak memory, written bytes and
anything else measured by Spawner.

1. Calculate standard solution resources for every test -- can be currently done by selecting last standard solution request, will also be useful for relative time limits.
2. Optionally, cache standard solution resources in tests.
3. (?) Add input_size field to tests, containing an arbitrary string
4. (?) Add more detailed keywords in o namespace, e.g. o.poly.n2, 'o.nlogn` etc.
5. (?) Add algorithmic_complexity field to problem_sources, assign it based on keywords and/or attributes in problem package.
6. Import some JS graphing library (also use it for console graphs)
7. On req_details page, add graph of resource/input_size
8. Add 'Resource' selector
9. Add 'input size source' selector of: explicit, checker comment, generator param + input file, guess by standard solution time
10. For the first three sources add sub-selector for integer value P. For each test, select P-th integer parsed from string -- so multi-parametric problems and imported problems with verbose checkers are automatically supported.
11. For "guess by standard time" add sub-selector or "algorithmic complexity", fill it with explicit data if available. Then, calculate approximate input size by fitting the complexity function to actual times.
12. Display calculated data on graph, with various visual controls, in particular linear/logarithmic axes, usage of test number instead of size, displaying several graphs at once.

So, i have the following proposals regarding UX.

I will update this issue ASAP.

I'm suggest to use following legend :
[D] - Declined
[A] - Accepted
[R] - Realised
[W] - In Work
Put it in start of suggestion.

Task page for authenticated users :

• Storing last user selected compiler on uploading task solve (as i promised i will implement this by myself)
• Select task by double click on it's row (it's really hard to reach radio button of that size)
• Will be nice if we have ability to upload task solve from other places (at least way to main system from task-view page, task page for contest for ex.)
• Drag-and-Drop?

Run details page :

• Definitely, "Compare" button shouldn't display for users, who don't have that ability
• Some information from other users runs display even if not jury, we should check that case completely
• By default, will be better if that page(from console for ex.) will open in new window, like this
• It will be nice, if we have option to watch testcase for failed test if some contest flag nor contest is not official(for ex.)
• Would be nice in case of runtime error or security violation display number of string that caused error, i think it is possible to do that

Console :

• Currently, it displays all runs for same task, why just not display last from that, or if needed make something like "spoiler"
• Auto-opening new run is useless, since modern browsers will block popup windows if them wasn't open with one of functions that originate from user event(click etc.), we can exclude domain from that behavior, but it's makes no sense

Results :

• Would be cool to see what language did used participants

Users of contest:

• It was really hard for me to locate pagination, intuitively it should be in down-right corner(why not just put a copy of it?)

Global update :

• Move everything to AJAX isn't bad idea, though
• We shouldn't display any system information, an example "Access Prohibited", "Incorrect.." blah blah, as like responses with error code 500, that definitely helps attacker detect system breaches
• Is json option in some places realy needed? If it realy needed, we should change src keys, so they will not equal keys in database, and return same information, that could be fetched from page if it was html
• Currently, current contest doesn't store(isn't it?), why doesn't store it?

Something can be overthinked, other can be lost.
Almost of them is pretty good concept.

User can't change his country on a settings page, no matter what country he choose from combo box.

This will lead to forget annual year changes in commits, like in af0d00c or e11ba13.

CATS возвращает кириллицу из JSON-файлов в неправильной кодировке.

В системе CATS указано название уже 1,5 года как несуществующей кафедры - "кафедры информатики ДВФУ".
Между прочим, на сайте http://imcs.dvgu.ru/works слова "кафедра информатики" выделены более крупным шрифтом, чем слова "математического и компьютерного моделирования".
Информатики "захватили власть" на кафедре!

Idea:
select count(*) from reqs where not_processed and sent_last_day and not unhandled

Optionally, since testsets are not necessarily continuous ranges.

Automatically add such test to the enclosing testset

It happens because the return code of git config remote.origin.url is non-zero and exec_or_die handle this as error.
@klenin

На главной странице CATS написано "модуль spawner". Для CATS это модуль, но для пользователя CATS это просто программа.
Это мелочь, но использование простых и понятных терминов - это тоже важно.

Similar to tests

Нужно предоставить интерфейс для получения кода засабмиченных попыток в формате JSON.

There is a button in the solution code viewer, placed nearly from the top right corner, with "?" char as a caption. This button looks so strange - when I press it, CATS pushes me a little window with the single "@ABOUT@" string in.
Is this button a something like unused feature of implemented code viewer control? Or this is a localization bug?

User expierence improvance for results page

• Would be cool to see what language did used participants

1. Add some problem to contest
2. Change letter to another letter
3. Replace doesn't work until page reloading

User expierence improvance for task page that can be seen by authenticated users

• Storing last user selected compiler on uploading task solve (as i promised i will implement this by myself)
• Select task by double click on it's row (it's really hard to reach radio button of that size)
• Will be nice if we have ability to upload task solve from other places (at least way to main system from task-view page, task page for contest for ex.)
• Drag-and-Drop?

User expierence improvance for run details page

• Definitely, "Compare" button shouldn't display for users, who don't have that ability
• Some information from other users runs display even if not jury, we should check that case completely
• By default, will be better if that page(from console for ex.) will open in new window, like this
• It will be nice, if we have option to watch testcase for failed test if some contest flag nor contest is not official(for ex.)
• Would be nice in case of runtime error or security violation display number of string that caused error, i think it is possible to do that

В ответе на запрос вида http://imcs.dvgu.ru/cats/main.pl?f=users;sid=;cid=;json=1; не возвращаются id пользователей. Они нужны для выбора подходящих попыток

http://imcs.dvfu.ru/cats/?lang=en
should be "...sity" instead of "...suty".

sorry if it's too minor issue.

The orange background laying under the contest name on top of the page do not cover all area it supposed to cover.

Extracted from issue #37

При нажатии на кнопку "переместить" открывается страница для выбора задачи. Если переключиться на другую страницу (я имею в виду нажатие на ссылку с номером, например, "2"), кнопка "переместить" меняется на кнопку "прилинковать".

Use '?' instead of traditional '+' or '-'

See 37cd8a8 for current implementation

Что, если сделать в CATS ещё одну разновидность турниров -- теоретические турниры, когда решением задачи является текст?
Текст можно либо вводить в поле ввода (оно в CATS уже есть), либо отправлять в виде файла (pdf или doc, например).
Жюри ставит оценки за решения, оценки отображаются в турнирной таблице.
И ещё, кстати, было бы неплохо добавить к системе CATS форум, в качестве форума можно использовать группу Google.
Ещё есть вот такой сайт: http://vk.com/mathmachine.
Хорошо бы использовать его в учебном процессе со студентами и школьниками.
Там, кстати, можно проводить творческие конкурсы, как на Физтех.Радио (http://vk.com/radiof).
Уже пробовали играть там в "Поле чудес" (на математические темы) и в игру типа "Кто хочет стать миллионером", когда участникам предлагаются опросы.
Этот сайт может также служить площадкой для различных обсуждений.
На сайте есть разные картинки с так называемой матмашиной и просто интересные картинки на математические темы.

Currently, textarea submissions for PascalABC and FreeBasic do not work because compilers are picky about source file extensions.

Не сравнивает ни у одного и того же, ни у разных.

Options of contest:

1. Score last run -- as is
2. Score best run
3. Score chosen run
4. Score chosen run with tokens (a-la ROI)

I suggest to use powerful github thing in 'Issues' called Labels, so if issue bout 'bug' we can assign 'bug' and etc. As you can see i had marked few last issues with '[UX]', labels is better way to do that.

I suggest for now :

• Features request
• Bug
• Security
• Improvance
• Discussion

В системе CATS проводится много различных тренировок, в которых потенциально может участвовать (официально либо вне конкурса) любой желающий. Участие в тренировке представляет собой, в первую очередь, «общение» с тестирующей системой CATS.
Я считаю, что для полноценного изучения информатики необходимо общение друг с другом и с преподавателем. Такое общение в очной форме организовать сложно, тем более что количество общающихся достаточно велико.
Отсюда вытекает следующий вывод: необходимо создать (разумеется, на базе системы CATS) среду для обсуждений и обмена опытом.
За основу можно, конечно, взять опыт проекта codeforces, но нам всё же нужен региональный проект для наших студентов и школьников.