Luke Gorrie

Getting there! Just pushed implementations of five more bytecodes. I am working on running the test suite like this: ``` raptorjit/testsuite/test$ ../../src/raptorjit test.lua ``` In practice this starts executing bytecodes...

More baby steps! Return to Lua function calls is working now. So are the bytecodes `CAT` (concatenate), `FNEW` (instantiate closure), `TNEW` (create new table), and `CALL` (function call from Lua.)...

(This reason that work on this branch is paused is that I'm taking time to refresh the Studio user-interface to make it more approachable. That's nearly done now.)

Thanks for the report! This problem seems to go away when merging the latest `lukego/raptorjit-upstream` branch. This could be (?) due to a new fix for a problem where RaptorJIT...

@kbara it would be awesome if you take the lead on this!

@wingo The main value I see in this activity is to ask ourselves, "How should Snabb application code really look?" Consider a really simple example like [example_spray.lua](https://github.com/SnabbCo/snabbswitch/blob/master/src%2Fprogram%2Fexample_spray%2Fexample_spray.lua): ``` lua module(...,...

[Torch](http://torch.ch/) is another LuaJIT application similar to Snabb Switch. Their [API documentation](https://github.com/torch/torch7/blob/master/README.md) is worth checking out, for example the [Tensor](https://github.com/torch/torch7/blob/master/doc/tensor.md) object. IIRC the `torch` module was the inspiration for having...

@sleinen Yeah :). I did ask their advice (via mailing list) before adopting the "busybox" structure of a unified executable that includes everything. They went the opposite direction and decomposed...

@eugeneia On the one hand I agree that the overall structure may be fine and having "more than one way to do it" for libraries is healthy. Even so I...

Further thought: The OO convention documented on #740 could also help to relieve verbosity. Procedural code accessing a `snabb` module might end up more verbose even without the `require()` calls:...