Accel_T v1.52 by Equalizer

[Archon_Wing]

Here's 2 reps a pv2t and a zv2t. I'll add comments after I watch them but one thing I noted is that the AI's attack forces seem kinda small later in the game, at least in the pv2t rep. It could also expand/attack early a little faster. The other thing is that wraiths are annoying, but aren't too good in the hands of a comp, so i don't really recommend them-- rather see another bc or two Defense was pretty good, and those Goliaths aren't too bad. Mines can be useful vs zerg late game.

Instead of 2 armories, I think the comp should focus on the infantry upgrades first since infantry is the bulk of its force. So maybe 2 ebays and 1 armory instead. Another thing is that I didn't see any turrets in the outer expos; might have been the map though.

[Archon_Wing]

Ok, Now that I have some time, I can go over your actual script. It's quite detailed and covers stuff I would have never thought of. It's also quite compact ~1000 lines which is very nice.

Suggestions:
Early Rax timing is a bit off. It builds rax too early vs hatch first and too late to defend vs early rush in time. Though the later requires a bunker or two.

But I think I may have a nice idea for this

--requeneBarracks1--
rush(0, accelerateBarracks1)
resources_jump(375, 0, 12cc)
wait(50)
goto(requeneBarracks1)

--12cc--
wait_buildstart(12, scv)
expand(expand1)
build(1, barracks, 80)
...
Will need adjustments, of course... but that's the general idea. I think it'll cover pretty much any zerg opening.
This enables the comp to do a very fast expansion if it has too many minerals and enemy has no pool, or get rax and bunker to defend against a rush. You can always add a random_jump to mix it up a bit. An bunker would be needed at the natural asap for obvious reasons.

Another thing of note is that I did not see farms_timing or a building of a supply depot til very late. That could cause some slowdowns. Supply building should start either after a marine starts building or the 9th scv, whichever comes first.

Gas is a bit too early for what you intend to do; causing a massive gas pileup. Usually 2nd rax would be built after first unless you intend for a early factory.

One possbility is rax---> gas---> factory---> tank---> academy--_> more barracks, for a really fast m+m tank push.

The last thing is that I think the expo loop is running too fast, causing it to spread itself too thin. Especially concerning the difficulty of terrans in defending their expos, I'd say a wait time of 8000-9000 may be more beneficial.

[Archon_Wing]

0.4x has been added to the database. I took a look at an AIvsAI rep and it does very well. I think it emulates the "standard" build better than any other script before has. If the script is set to attack with a science vessel or upon hive detection, it'd be pretty strong and truly replace the blizz ai. I'll post reps when I have the time.

[Myk]

[Equalizer]

Regarding overloading, (the wait_force(255, [military unit]) command) I've tried it a few times with the train command and it doesn't seem to work the same.

Overloading doesn't "run out" however it does some times "get lost". In that if there are other multiruns called, just like any other multirun the script handles, these can loose enough priorety to stop being read. I've been experimenting with recalling the important ones (marines and tanks) and other ways to make other unit types still get used even if the overloading gets lost.

I really am unsure if target_expansion() actually works but whether it's the clear clear_combatdata() or it or somehow the two working together the AI seems to hit expansions enough to say that together they have some effect.

About defaultbuild_off(), I am not completely sure of all the implications of default build but I found without calling defaultbuild_off() I had issues with the AI liking goliaths too much.

The version I've been currently working on is starting to look less like the default AI in the way it plays and more like... umm well it likes marines and tanks a lot and attacks more and will through what ever else it feels like into the mix.
It still is unfortunatly unstable though, like said above overloading isn't the perfect fix to unit production.

[Archon_Wing]

[wscha231]

I'm so impressed with accelT.
0.7 looks really strong in late game.
however, accel is usually defeated by early rush such as 4 zealots, dragoon push, and speed up zerlging.
I hope you work on early defense. you may let ai build bunker in expansion .

[Equalizer]

I have yet to have the AI properly counter Protoss (all triggering is zerg specific so far) so I am not surprised that protoss can easily beat it early game.

As for zergling rushes I've tried to get the AI to catch early pool and change build order.
It seemed to do reasonable againts other AI ling rushes (Unleaded's 5 pool and Panda's two base zergling rush).
I'd like to know which of the 3 possibilities (really early rush eg. 4pool 5pool etc., one based zerg ling rush, two based zerg ling rush) version 0.7x was loosing to so I can look at what it could do differently.

Update: version 0.8x has modified start build that should be somewhat more resistant to rushes

[Archon_Wing]

The rush detection is really good. I was thinking about adding it to my own scripts and maybe even as an entropy master terran opening if that is ok with you.

The targeting expansion is a command I've never seen before, but it works well.