If you do not at least have a tool to feed the context of the LLM then you will always have a b*tch of an experience, I believe. We tried out an LLM in our company which had almost no context management. It sucked. Everyone hated it.
Queue copilot (which is arguably likelty not the ultimate toolchain) which can at least read files it needs etc, either explicitly or implicitly, is a whole other ballgame and it works 100x better.
Author here. Haha that's so creative! I definitely thought about lowering the limit, but then it wouldn't really be ASCII-esque lol. I might just lower it so that you don't have to do this workaround.
This sounds like a flashback to J2EE. Which I know is still alive and well. Banks, insurance companies and the tax agency do not much care for fancy new stuff, but that it works.
I describe these techs like garbage trucks. No one likes to see them but they’re there every day doing a decent part of what it takes to hold society together hah.
I dont understand, the agent mode of copilot will search for and be pretty good and filling its own context afaik. I never really feed any of our 100k+ lines legacy codebase explicitly to the LLM.
I really need this in my life. Once upon a time, things were good and our Chromecast with Google TV knew _exactly_ how to turn on our soundbar, set our TV to output sound to said soundbar, control the volume on that soundbar using IR.
Now absolutely nothing of that works. The audio output on the TV is set seemingly semi-randomly depending on content!?. The volume controls just stopped working, and I can not FIND THE SETTINGS in the menus? I suspect it is required to completely redo the remote setup to see those settings, OR as I rather suspect: they broke this shit in purpose to get us to buy a new Google TV Streamer.
I feel very strongly after 20+ years of development that DRY is a good guideline, but I have also seen many, many times that trying to follow it to the letter is actually detrimental and results in too complex solutions.
This is one area I do use LLMs for, write small utils and test code, and very often in languages I very seldom even touch, such as C# and Rust. This to me seems to be the core idea of a tool that is awesome at translating!
Is it just me, or should they have just reverted instead of making _another_ change as a result of the first one?
ALSO, very very weird that they had not caught this seemingly obvious bug in proxy buffer size handling. This points to that the change nr 2, done in "reactive" mode to change nr 1 that broke shit, HAD NOT BEEN TESTED AT ALL! Which is the core reason they should never have deployed that, but rather revert to a known good state, then test BOTH changes combined.
Queue copilot (which is arguably likelty not the ultimate toolchain) which can at least read files it needs etc, either explicitly or implicitly, is a whole other ballgame and it works 100x better.
reply