Hi Neil, On 28/11/2005, at 9:23 PM, Neil Mitchell wrote:
Hi Tony,
On the Haskell IRC [http://www.haskell.org/hawiki/HaskellIrcChannel] a user named stepcut mentioned they were thinking about porting yhc to PalmOS, and I think they did some initial work. Unfortunately I don't have a real world address/name for them.
From what I remember, the issues they mentioned to me were:
* No support for a 64bit long at all * Lack of GMP for PalmOS * No real files
These sounds familiar from my nhc runtime port work. My take on the first and second are that typical handheld apps are not likely to need these features (or can work around them) so a handheld runtime can omit them (at least at first). The third is possible using stream databases and I already have some code for the nhc runtime that may be usable in a yhc runtime port.
So, my question to you is: Would you advise active use of the yhc code as a basis for our handheld work? Is it stable enough to be useful for this at this point? We can probably put up with deficiencies at the compiler end since we don't expect to write very large or complex programs in the near future but we need the runtime to be stable so we can be sure of the port. Or should we wait? Any other comments?
The compiler is under active rewrite, and is not particularly stable - but should be sufficient for what you need. The runtime is much more stable - almost all of the changes in the last few weeks have been minor and as a result of ports, to make it more cross platform. A port should not require porting the compiler much, if at all - when the compiler is self-compiling (only minor issues away from this, nothing particularly hard) - it should run on top of the runtime.
Ok, thanks. I'll definitely look into it and report back when I have something useful to say.
One thing in Yhc's favour is that the code is quite clean and standards compliant, it was rewritten from the ground up quite recently with the goal of portability in mind.
Indeed, this is potentially a big advantage. Thanks, Tony
Thanks
Neil