Not working with secondary source of prompt information
Sukaritat opened this issue · 2 comments
BTW. The behavior I describe below is consistent with 2.6 and with a manual install of 2.6.1
Since installing Talk to Chat I have used it concurrently with Chat Prompt Genius which allows me to keep a repository of prewritten prompts that I can choose among and populate into the ChatGPT enter box by selecting from the list of saved prompts I have stored when I type in the "/" After hitting backslash I select the prompt I want to use. All of my pre-written prompts variations on a et of instructions regarding what I want Chatgpt to do or rather how I want it to process the information that I am about to give it and that I want it to consider all within the same question/prompt. Up until a very recent change, either by Talk to Chat or by Open AI, I have been able to insert the prompt I selected, move my cursor to the end of the text where the last characters of the prompt appeared in the "chat input box" and then use my voice to add additional text to the still open prompt using the Talk to Chat voice command. Only when I had finished adding all of the information that I wanted Chatgpt to "process" according to the instructions I had given via the insertion of the prewritten command did I hit "submit" or "enter" either manually or with the verbal command of "send message now" at which point al of the information, the pre-written prompt and the additional information added via my voice be submitted.
Now, when I attempt to do what I had been doing above, the prewritten commands will continue to be populated in the inbox when I select the one I want after pulling up my choices after having entered the "/" However, now when I attempt to add any additional information using my voice, the microphone will not pick up any sounds, nor will any additional dictation be added to the currently open prompt. It is as if the application has frozen. Worse if I do get it to open up once it comes online it instantly erases over the pre-written prompt text that had been in the "input box". What mess did Open AI create with this seemingly innocent addition of the color green to the "enter" button. Everything had been going so well.
Hi,
I think that OpenAI in their latest update has broken a lot of extensions like mine and probably Chat Prompt Genius.
Once they figure out how to fix it (as I did in v2.6.1) things may work properly again?
The fix in version 2.6.1 is actually really simple and didn't change much of the behavior.
It just sets the textarea's value in a more 'native' way, simulating actual key presses. Which hopefully won't be preventable by OpenAI.
As for the code it's pretty simple:
When a message needs to be added to the prompt, TTGPT checks whether there is already text in the box. If there is, the new text is appended to the box. If there isn't, the new text is just set as is. This part of the code hasn't changed.
Thanks for the quick response and for the clarification that the problem needs to be corrected on their end. I will knock on their door if the behavior continues.
Btw I find your extension STILL to be the best (by far) in this class, however, I do wonder what happened to the ability of the app to display a "PAUSED" indicator at the bottom of the Talk to Chat GPT controls/settings box that appears in the top right corner of the screen when the listening function of the microphone was paused. Without it I find myself having to constantly toggle the mute function of the microphone on and off for me to understand if the app is on, first of all, and secondly, to force the listening bar to transition from the color grey to green and then finally to red, at which point the microphone is usually, but not always, in the proper listening mode. Perhaps it is just my set up, but things seemed to function more smoothly and as designed and described (in earlier releases) when the "pause" status was visible on the controller. It seems the pause indicator function may have been taken over by a chrome browser pop-up that appears indiscriminately, based on my first impressions, and which contains a status message that I find to be confusing insofar as it seems to describe app behavior that is inconsistent with how the app was described as behaving in earlier versions when it was toggled into the pause mode. Moreover I am not sure if it is consistent with the setting choice that currently exists in the latest version about the microphone, listening status and pause mode. Perhaps the next version can provide some clarification about what it means for the app to have the microphone on or off and listening or not and how those modes are controlled by and visible to the user.
Lastly, Am I misremembering or did there used to be a setting that told the app to always refrain from automatically reading the chat's responses out loud thereby eliminating the need for the user to hit the > button each time to stop it? If it is another extension that I tried at one point that had this feature and not yours, I apologize, but it would be a welcome feature, fyi.
(PS I hope the above is received as an informative observation and not a complaint from the ungrateful)