After we interviewed Igor Stasenko about NativeBoost – a FFI implementation for Pharo/Squeak I took a closer look at it and looked for projects testing it.
I found two projects I actually would need in the future: 0MQ as a communication library (make it possible to use Pharo within our IT structure) and ICU for having full Unicode/Timezone and more support.
There was initial source code for ICU from Jan van den Sandt and I looked how to use the source code from my VASmalltalk port of ICU under Squeak/Pharo.
The result was, that I port the OS* class structure from VASmalltalk to Squeak/Pharo, put this on top of NativeBoost and then tried to use the code from the VASmalltalk implementation. After all: NativeBoost does its job. Perhaps here and there some more detailed error descriptions would fine – and some questions about deep internals are still not answered. But it will work.
The code is actually not public yet, because it is not finished at all and due to the fact, that I make a rewrite and refeactoring (at the same time on both platforms) it takes again more time than I initially thought. But it will be made public, when at least a suitable subsystem is ready to use.
At least today, the instance of the MSKLocale (which will be renamed to ICUSystemLocale) is now initialized correctly and contains information about the locale, timestamp, timezone, printing format and all that stuff.
Then some words about the latest version at vastgoodies.com – I actually published it a little bit too early. It contains experimental code to find out, if a Regex implementation could be created by using the ICU Unicode regex machine – but it is not ready and lots of the test cases are still failing.