You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Sep 27, 2024. It is now read-only.
In the rich text editor, we start out with an empty composer. If the user inserts a new paragraph, then removes a paragraph, although the composer looks empty, it in fact contains a single p tag. This changes the output in the message body as shown.
steps
model
start composer
""
insert command
"/sample_command"
remove all text
""
insert a paragraph
"<p> </p><p> </p>"
remove a paragraph
"<p> </p>"
insert command
"<p>/sample_command</p>"
This means that if a user sends a slash command after these steps, it is incorrectly recognised (due to the opening of the p tag) and incorrectly processed. The slash commands seems to be designed to work with a string of text arguments after the command, so we could simply strip the opening and closing p tags from a message before we send it.
This could also potentially be rolled into the task #679, as we could simply output a raw text string if the rust model detects that the message starts with a command. I think this would be the neatest solution.
The text was updated successfully, but these errors were encountered:
In the rich text editor, we start out with an empty composer. If the user inserts a new paragraph, then removes a paragraph, although the composer looks empty, it in fact contains a single p tag. This changes the output in the message body as shown.
""
"/sample_command"
""
"<p> </p><p> </p>"
"<p> </p>"
"<p>/sample_command</p>"
This means that if a user sends a slash command after these steps, it is incorrectly recognised (due to the opening of the
p
tag) and incorrectly processed. The slash commands seems to be designed to work with a string of text arguments after the command, so we could simply strip the opening and closing p tags from a message before we send it.This could also potentially be rolled into the task #679, as we could simply output a raw text string if the rust model detects that the message starts with a command. I think this would be the neatest solution.
The text was updated successfully, but these errors were encountered: