Re: Lanes and Lindas - 2008 revise

[Irayo <irayo.lt@...>]

Asko Kauppi wrote:
> 
> These issues are listed in TODO and BUGS.
> 
> 'make basic' works, and can be used for proofing the system on any
> platform.  I will have a look at the x64 things you reported.
> 

Oh wow, I'm sorry, I didn't think to check the known bugs list :)  I
feel silly now.

In that case, `make basic` works properly 100% of the time on my x86
Linux, but only succeeds about 85-90% of the time on x86_64.

About the sublanes, I was referring to having lanes creating lanes, e.g.
recursion and stuff.  For example, I have a tiered system at the moment:


dispatcher +->   connection +->  socket receiver thread
           |                \->  several data handling threads
           |
           +->   connection +->  socket receiver thread
           |                \->  several data handling threads
           |
           \->   connection +->  socket receiver thread
                            \->  several data handling threads

Sorry for my crude ASCII art diagram, I'm in a hurry :)

The dispatcher only creates the connection lanes; the connections then
create their own (sub-)lanes for socket receiving, data handling, etc.
In some cases the data handling lanes may even need to create a third
tier for database access, etc.  I can't seem to figure out how to do
this from the docs, is it possible?  Everything I've tried either
segfaults, hangs, or throws nil reference errors.  I've glanced at the
tests as well but I don't think any of them show this sort of structure.

-- 
Irayo