xorian: welcome back. xorian: I think I've got those changes ready to go. Any preferences as to what I call the branch? xorian: (s) scottven: Are we talking about just the changes in basics/basic and basics/generics that we made last Friday? Or did you do some work in vesta/cache too since then? xorian: cache too scottven: For the basics libraries, I might append ".atom_no_gc" for the branch name, and for the cache maybe ".client_no_gc" scottven: Feel free to make your own choice though. xorian: ok. I think I need to do a bit more work in cache yet before it's ready to check in, but basics & generics should be ready to go. I'll try to get them into branches tonight. wishes he could /**nocache**/ a _run_tool() call. batsonb: Usually a _run_tool is a significant enough chunk of work to be worth caching. Not that I haven't seen some really small tool runs. batsonb: So, why do wish that in this case? xorian: I have lots of cases where I have a _run_tool() call sitting inside another function, and the function does some nontrivial work before and after the _run_tool() call. xorian: so, I'd like to cache at the parent function granularity, but then I get two cache lookups everytime the function misses. batsonb: I'm guessing it's unlikely that the function could miss and the _run_tool hit? xorian: right, very unlikely. xorian: also, with functions we have the handy /**waitdup**/ pragma, but I think one can only use that if the function is being 'cached', correct? batsonb: Is the outer function doing other cache-able things (calling otehr functions)? Or is it all inside the function or /**nocache**/ functions except for the _run_tool? batsonb: Correct, /**waitdup**/ wouldn't do anything for /**nocache**/ functions. The /**waitdup**/ machinery is part of the caching machinery. xorian: it's doing a mix of calls to other cache-able things and local work. xorian: So, that's another reason, the /**waitdup**/ primitive is quite important for trimming the number of cache lookups in my build, but it only works if caching is enabled for that function. batsonb: Well, it's of course not like it would be impossible to implement that. batsonb: I'm not sure how soon I might happen to get to doing that though. Honestly, it was a little easier for me to make a case for spending time on the "Waiting on possibly identical..." thread count issue, as that clearly has some benefit for us here as well. xorian: Understood--just a nice-to-have. If you think of an easy way to do it, let me know and I might give it a shot. xorian: ok. 3 branches, all checked in.