Got the idea while figuring out how to squash commits in Git Bash.

The following do the same thing:

git rebase -i 60042c8

git rebase --interactive 60042c8

My idea is to apply this same concept to commands we use with the bot. The (obvious) benefit is being able to reduce the typing required for commands while preserving the longer, more descriptive command names as well.

Example:

$ role add Junior

Could also be called using

$ -r -a Junior

or something like that.

Ideally we'd want to agree on a naming system for the abbreviations

Would tie in well with updating the help command (#127)

Something cool would be if the bot had a way to play poker with other people. Start each player off with like, 100 coins, and they could place their bets through a DM message and win/lose coins by playing poker.

$poker [<@User1>, <@User2>, ...<@UserN>]

This command will depend on the upcoming designated roles feature so as of rn its unfinished

I noticed that several channels on the Discord could use a simple calculator as those classes are heavier in math. What specific features and functionality would you like to see?

this is core bot functionality, we need a way to manage classes and roles without having to do it manually, there is a decent amount that will go into this database and service wise so this will be a rolling pr

Currently some exceptions are swallowed at the top level, we need a way to bubble them up to discord to be logged

Right now all references stay in the reference cache regardless of wether or not they are dead, they should be removed from the messenger if the weak ref is no longer alive.

Any ideas of what it should support.

Currently the db layer doesnt check before it attempts to insert a record, this needs to be changed and a pre check inserted. Exception control flow is never good

Pretty self explanitory. If the user sends a command that doesn't exist, e.g. $dice_role, the bot sends with the error message a list of roles that do exist. Ties in with the changes for #127.

Additional Idea:

If the user sends an agrument that doesn't exist, such as $role new, the bot could provide a list of acceptable arguments.

Problem:
Help message for arguments could be problematic. For some commands the user can provide input, such as $role requested_role. I think an additional argument would need to be added so the bot knows to not treat the user input as an incorrect argument. (e.g. $role request requested_role)

The bot will give you a random trivia question and then you can attempt to answer it

This is a great feature as it can be as simple or as complex as the implementer would like. and there are many ways to tackle this. This would be a great first issue for someone to get their feet wet with discord bots

Better exception handling that's less obvious plz.

Typing in currency shouldn't spam the chat with errors

Dynamic role grouping to set per cog perms on a per server basis should be implemented

There was some talk about a TLDR command/channel being made.
Some ideas that were brought up.

• Taking a link from the message at the beginning of a conversation. Adding a TLDR for the conversation. Either adding the TLDR and link to a TLDR channel or making it searchable, maybe by title.
• Finding some way to link messages together and adding them to a TLDR. This could take upvotes for what is important to the conversation in to account.
• Calling tldr with no arguments returns the most recently created TLDR

The default d.py help command is ugly and a way to make it pretty would be very nice

Currently not all actions the bot takes are logged, this issue will serve a center point for log statement related commits to go. This is a great way to get your feet wet with contributing to the bot and with githubs workflow.

We should have a utility that allows for a user to react with an emote that tells ClemBot to delete its message. To avoid cluttering up the channels

PyPubSub is synchronous and not async which makes it impossible to use in this context. Upon research it appears that a light weight asynchronous event aggregator. the only thing i have found is this https://pyzmq.readthedocs.io/en/latest/api/zmq.asyncio.html.

TODO: to move forward we will need to either roll our own event framework or use zmq. if anyone has any further suggestions or comments please leave them down below. We also need to strip out the pypubsub dependency and generate a new freeze file

The alias "set" would work work really well on the !channels add command

We need a way to idiomatically define how services handle their startup and shutdown procedures, i propose an abstract base class that defines a load method that handles such things

match students in Clemson School of Computing that have similar hobbies and classes using a nearest neighbors search

This is pretty simple,

when posting a comment link in the channel, clembot replies with the message and some details

when adding a newline character immediately following the message link, clembot will not link the comment details.

the grades command has been very popular and helpful. However, it is often wanted to compare individual teachers to make an informed decision on who to take. I propose if possible we add a way to get professor specific info

Having a way to flip a coin in discord could be cool.

Small bug I noticed when asking the source command to print specific lines of code. The last line seems to get dropped.

Example <bot_prefix>source example_cog.py 1 3 would only print lines 1 and 2:

1:    #what the bot does 
2:    import logging

we should support the ability to have the tag be inlined in a message so if the tag is $world

then the message hello $world should trigger the accompanying tag

Currently the messaging framework just awaits the events but in order to be a true fire and forget the events will need to be wrapped in a task and used that way.

Since Jay is a coward, only the first tag in a message will work correctly:

this first $tag1 will work but this second $tag2 will not

Obviously, this needs to be fixed.

Testing webhook

Currently the logs overwrite themselves on restart. This is bad practice as history is lost. I suggest we move to a date specified format where each restart takes the current time and appends it to the file name, making it unique to that run

The !crab command saves the .gif file with a filename that contains invalid characters when the bot is running on Windows.
The datetime.datetime.utcnow() needs to be formatted with strftime()to remove colons (and maybe spaces?) instead of being used directly in a format string.

We need to chunk the edit and send it that way

Currently the dynamic loading code only searches the current layer in the cogs and services folder. this makes it impossible to have nested cog/service packages. The dynamic loading code will need to be rewritten with this in mind.

Currently you can only assign one role at a time, eg:
!roles cpsc-4910
!roles senior

Instead it would be nice if users could self-assign multiple roles at once, eg:
!roles cpsc-4910 senior etc.

Should be pretty self explanitory.

Normal Example
Roles:

1. Freshman
2. Sophomore
   ...

User: !role sophomore

Bot: Role assigned: Sophomore

Special Case Example
Roles:

1. JUNIOR
2. junior

Ignoring case, there are two possible roles the user is requesting. In this situation, my thought is to perform a case-sentitive evaluation of the user input corresponding to the roles. If not role exists, return an error listing the possible roles the users might've been requesting. Since there are two possible roles the user is requesting, the bot will need to

Special case example 1:
User: !role Junior
Bot: Error, role not found. Two possible roles: 1. JUNIOR or 2. junior

Emojis could be used to have the user select which role they wanted.

Special case example 2:
User: !role junior
Bot: Role assigned: junior

For this special case, I'm wondering if there should be a warning even if the user types in the case-senseitive role to inform them of another possible role they might otherwise be unaware of.

An authentication module is badly needed as the default d.py auth code is very very limited

The message should be validated before the bot tries to build the embed

when in a channel with an associated role (from the class cog), the command (i.e. !role apply) would add the role for that class to the user who called the command.

For example: When I am in the cpsc-1010-teacher channel and run the command !role apply, I would receive the cpsc-1010 role

SoRt Of LiKe ThIs FoR sArCaSm

A cool minigame feature would fit in well in the server. This is also a great way to dip you hand into the database layer

Currently we are using $ due to a conflict with ! but any other ideas are welcome

Hello,
In the channel that uses this bot, we had considered the Slack feature of threading comments.

To do so, you must leverage the properties of a thread of messages - Also called a tree.

For example (messages are numbered for use later, These numbers will be used in place of links),
1 Student 1: I have this question.
2 TA: This is the response to your question
3 Student 1: Please clarify this point
4 TA: Here is clarification
5 Student 1: $Thread_close

When the $thread_close command is issued, the bot will be able to do the following
Take the link to last message in the thread
put the link on a stack
navigate to the link on top of the stack
If the new current message has a link, put it on a stack and recurse. Otherwise, pop the top off the stack and append the message to the transcript.

Using the example conversation above, the bot will construct the stack until it reaches a message with no link.
The stack will look like this (top) {1,2,3,4} (bottom)
the bot will then pop the top, follow link, and append the contents to the transcript until the stack is empty.

Thus the transcript becomes (contents omitted, messages referred to by number from original conversation)
1
2
3
4

This allows for each individual offshoot of a thread to be closed independently. In the context of a tree, this means a leaf node can either spawn a child (a reply to a message in a thread) or decide to stay a leaf (close the thread). Because the action of any leaf is independent of any other leaf, every thread can be closed and have its path from root->the aforementioned leaf node.

For another example - Consider Students 1,2,3 and the TA named "TA". They may have a conversation like so
1: I have a question
2: <Link to 1's question> That reminds me of my question
3: <Link to 1's question> This reminds me of a different question.
TA: <link to 1's last message> Answer to your question
TA: <link to 2's last message> Answer to that question
TA: <Link to 3's last message> Answer to another question
1: $Thread_close <link to the TA's answer to their question>
2: $Thread_Close <link to TA's answer to their question>
3: <Link to TA's answer to their question> Please clarify on your point.
TA: <Link to 3's last message> Here is clarification
3: $Thread_close <link to TA's answer to their request for clarification>.

Thus a tree is created. In the tree, the links were replaced with edges from parent to child. To see the tree, please view the image here: https://i.imgur.com/zw1Opop.jpg

This is each individual transcript that individual users would see
1 -
1: I have a question
TA: <link to 1's last message> Answer to your question
1: $Thread_close <link to the TA's answer to their question>

2 -
1: I have a question
2: <Link to 1's question> That reminds me of my question
TA: <link to 2's last message> Answer to that question
2: $Thread_Close <link to TA's answer to their question>

3 -
1: I have a question
3: <Link to 1's question> This reminds me of a different question.
TA: <Link to 3's last message> Answer to another question
3: <Link to TA's answer to their question> Please clarify on your point.
TA: <Link to 3's last message> Here is clarification
3: $Thread_close <link to TA's answer to their request for clarification>.

I believe I have belabored my point by now. If you have any questions please place them in the comments.

I envision a command that you can call that will send back a random gif, will probably need to be filtered for nsfw results