The target list has been supplanted by PerfectTargets (albeit minus the non-tank names) and the aggro notifier has been replaced by Squishy, but I have never found another mod with the "SmartActions" feature, wherein SPELLCAST_START is hooked to automatically pick a target if none exist, or to assist a friendly if they are targetted. It's insanely handy for we button-mashing caster types.
As SA exists now, the actions work relatively well (there's still a bug where some non-spells don't work, i.e. Shoot Wand) but the whole addon is bloaty, and if one leaves the list enabled, it churns through 25 KiB/sec (!!!) in a full raid.
Additionally, its assist logic currently only supports CTRA tank tables (though adding oRA would be trivial), and the "marked target" bit only works if the list is on.
What I'd really love to see is a lightweight (ace or otherwise) rewrite of the assist logic that (optionally of course) uses PerfectTargets to assist based on arbitrary criteria (i.e., is my "assigned puller's" target, is marked, has the most people targetting, has no oRA/CT tanks targetting it, etc) both with an assist key and optionally with the automatic assist logic.
I have to agree that the addon does consume VM memory at too high rate, but is it really much of problem? When VM memory starts to be low it starts garbage collector and frees all memory just fine. SmartAssist does NOT leak memory. Only downside with the mod is that garbagecollector runs more often and when that happends it causes small (~1 sec) pause in the game. On my 70+ addon setup that happends only once every hour or so. The memory problem resolves automaticly when TBC comes with better garbage collector, or if I see it fit to use compost to ease the GC problem ... only question is, is it worth doing for few months.
But more importantly I'd like to know what makes you think SmartAssist is a bloat? The main mod including target list generation + assist logic takes only 2500 lines of code. Is it the amount of options in the configuration? Most of the options there are only additional condition in IF sentence, I'd estimate that dropping all "useless" features would reduce the codebase only about 300 lines. Same goes for using Ace2 lib, there is not that much usefull functionality in it. The configuration screen is likelly the most bloated one due the number of widgets there. And that is currently Load On Demand addon so it's not even loaded usually.
Instead of calling out for someone to ripoff my good ideas (which you call "Aced") I would prefer constructive critism & feedback. I can drop off features if someone manages to convince me that they are useless. I have done it in the past.
Codebase is small, clean, mostly kludge free and efficient (except the memory usage, which is kinda LUAs fault). I don't see any reason for calling "rewrite".
Next major version (1.5.0) contains theme support, making the addon look really good and extremely customizable. I've been planning to add also perfectRaid stylized target list for those who are not running game in very high res or prefer smaller list.
Wow good to see you here mate, i have been using this addon since well as long as i can remember. all I can say is that i would love for wands to be fixed and some other actions don't automatically cast sometimes. I'll try to see what i can do to help ya hehe.
I have a good hunch why auto cast on assist might not have fired always, it checks before casting that the target is in combat. But when mob is pulled, game doesn't flag it immediattely in combat. The latest beta has modified this a bit so it will consider mobs that have someone targetted to be in combat. This is in my experience is more reliable than just using checking UnitAffectingCombat.
I have also TRAC on my site (http://addiktit.net/~paranoidi/trac/smartassist/). Feel free to create tickets if you have any issues or good ideas. You can also check what I've had in mind for the future from view tickets -> view by milestone.
And if someone wants to take part in development I can give SVN access aswell. Or you can submit patches as tickets :)
Don't get me wrong, SA is a great mod, there's just some more room for improvement. With both SmartAssist and PerfectTargets running, SA's target list churns 25KiB/sec in a full raid. PT is near zero.
So, I run with PT on and SA's list off. It's a workable compromise. I was hoping that the assist logic (and the SmartActions hooks) from SA could be adapted to use oRA(2) and/or PT instead of CTRA and its built-in target list.
Hmmh, I'll see if I can implement target list generation using CompostLib so the memory "usage" rate would be around zero aswell.
Perhaps then you can disable whole useless CTRA MainTanks display like I have ;)
Btw, when you disable Assist List (In 1.5.0 just Targets) it will change the assit order a bit since the asisting logic uses the data that list generation gathers (for example priorizing marked/depriorizing CCED).
Submitted version that contains Ace2 Compost2 which helps memory problems a lot. I was able to dropdown the "usage" from 50 KiB/s -> ~7 KiB/s (simulated situation, real life raid situations still unclear). I don't think there is much room or reason for improvement before TBC comes to the rescue with Lua 5.1 :)
Rev 325 contains this and many other changes, it is however still a beta. If you find problems with it, make a ticket :)
So barring the upcoming changes in TBC, I assume this mod is going to evolve into more of a "Focus setting" addon instead of a ins-combat taget chooser? This focus unit sounds damn cool, I hope some nice code is put out to take advantage of it :)
Love your addon and the new praid theme! And yes, I noticed my KiB/s was well under 20 using the latest test revision. It used to hover around in the 50's with earlier versions + all the rest of my addons running.
Do you still plan on adding Clique support later on?
Yes, unfornattely most of the functionality is lost cause. Sadly my effort to keep the addon as pure-intentational as possible didn't mean anything in the end. I never optimized the target selection based on target properties (such as health).
But I'll restore what can be restored as well as it's possible under new circumstances.
But before that I'll make fast resurrection without implementing anything new (1.9.0). Just get rid of everything that was broken and then start figuring how I could re-implement the lost features (2.0.0).
the most important part in this addon was the list of available targets. If I click a button or use the mouse is not that important in my opinion.
So if it is not possible to hit a button to assist one target in the list, lets do it manuell. That is even better then search your target in the raid with 40 people at one spot and 5 to 10 mobs arround.
you cannot assign click actions to dynamically modified frames in the new 2.0 secure UI. main tank lists like in CT/oRA still work since it's a pre-existing frame (though you cannot move or change them during combat lockdown). perfecttargets suffers from this same limitation.
blizzard has done a great deal to limit the use of addons to perform any kind of automation. Maybe they could be convinced to add something clever like "if I have no target, assist the base raid interface's main tank".. though i would imagine that's still too much automation for their taste.
p.s. thanks again for the work on SA, paranoidi. as I reread my original post I realize it's a bit critical - your addon is/was a great help.