@ponapalt Some quick and obvious questions to differentiate between the advantages and disadvantages of gpt and gemini: 1. there are three people A/B/C on the playground. a is playing tennis with someone, b is jumping rope, what is c doing? 2. Can a person survive if he has only one heart left? How can we protect our heart? 3. draw a beautiful woman 4. Write two sets of words, each with the same first letter and length in ascending order. 5. Translate some niche classical Japanese texts into other languages
@ponapalt I asked math questions, logic questions with traps, uploaded pictures of physics problems and asked about the correctness of the solution process **all in English**, and in general the gmini ultra was inferior to the gpt4 in every way, except for the pic-reading function, gemini may have called the google lens :blobneutral:
@ponapalt There shouldn't be a lot of actual math cacl in yaya, maybe we can try to get some ghost to use the actual bigint yaya first to see how much of a resource hog it is.
@ponapalt As far as my current knowledge goes just have the following: 1. image input and image output 2. allow shiori to read and set data via sakura scripts 3. maybe allow the dll to provide certain messages in each message to shiori.
@ponapalt If possible I'd like to like shioris, further split the existing logic of how ssp reads shell images and renders the shell as a dll file, so that the new ghost can perhaps just ditch the old shell organisation in favour of 3D models or live2d. For old compatibility, ssp will default to using the This split outed provider dll if the shell doesn't specify one. The dll file can call the dll file so no matter there is no problem how many layers of extra processing are added! And given that the splitting process will inevitably lead to conclusions about all the currently predictable conditions, this customised shell rendering provider specification will be the best fit.
@ponapalt Just a rough idea, is it possible to get ghost to allow some manipulation of the image in the final shell render, like non-geometric stretching and shrinking or something? I was just thinking about ssp when I was suggesting some games, and sometimes it's not practical to do multiple boob splits for dozens of shells in order to allow ghost to have optional boob sizes, so why not just morph the boob part of the final image? (This was just a spur-of-the-moment suggestion! It doesn't appeal to me at all as much as the animation script for the icons so it's not a necessity for me)
@sxi_k If you want, try Firefox or Firefox Dev, as far as I know Firefox hasn't encountered any related problems. :blobpensive: Though sometimes YouTube and Google pages will intentionally sleep for a few seconds when they detect they're in Firefox to make you think Firefox is lagging, but that can be fixed with a plugin that disguises Firefox as Chrome.😂
You could try f12 in the page and check the js error log, that might work Generally speaking the likely cause is that the browser is intercepting jsstp requests to the local port for security reasons, adding a whitelist or changing the relevant settings may fix this.
@sxi_k It is also possible that ssp is prohibited from creating http ports, so also restart ssp and check sstp related settings as well! :meow_thinking:
@ponapalt maybe we can do an EQUAL with the final version and if it's equal we can show that "there is not need to be updated!" or something. :blobok: btw, the ghost page(https://ssp.shillest.net/ghost.html) can be used to show whether ghost has been installed or not by using this
@ponapalt codes for jsstp 3.0.0.1: async function GetSSPVer() { return jsstp.EXECUTE({ "Command": "GetShortVersion" }).then( info=>info.unknown_lines[1] ).catch( e=>'unknown' ) } async function IsSSPrunning() { return jsstp.available() } I'll update jsstp in the future to have getting the SSP version built in as a method somename, but here's the current replacement code. It works, just a little ugly. :blobpensive:
@ponapalt@tukinami_seika I'm sorry, it's indeed a problem with my environment configuration. The `/d` parameter to start only specifies the directory where the program is to be run, and in the absence of a path it will still look for the executable in the environment path, so there was a problem with the way my cmd script was written (it was just running the Single-step profile in a different place).
I changed the script to @start /d %~dp0 %~dp0ssp.exe %* @exit /b %ERRORLEVEL%
Now it works fine! (As to why the script would cause the shell to hang until ssp ended if without using start, I'm inclined to explain it as a metaphysical problem and won't pursue it any further.) I'm sorry for the trouble I've caused you guys. :blobpray:
@ponapalt I used 【start /d "E:\ssp\" ssp.exe /g Taromati2】 in cmd to no avail! Must 【E:\ssp\ssp.exe /g Taromati2】 to work Why is this? Can you fix it? :meow_thinking: