Re: Strange etrace

On 02/02/11 02:02, Bill Frantz wrote:
> On 2/1/11 at 1:51 PM, [email protected] (Thomas Leonard) wrote:
>
>> On 1 February 2011 06:39, Bill Frantz<[email protected]>  wrote:
>>> I was running my data analysis program (Reads a file, processes
>>> the data and writes a file. No promises, single vat, single
>>> turn.) and I received the following etrace:
>>>
>>> === 2011-01-31T07:18:11.651Z
>>> (PendingEvent.report:PendingEvent.java:107) WRN
>>> causality: Problem in turn<<Vat start in<runs in start>>,3>:
>>> --vvvv--
>>> Failed: OldFarRef may only be smashed:<Far ref>
>>
>> There were a lot of problems like this in 0.9.2:
>>
>> http://www.eros-os.org/pipermail/e-lang/2009-November/013314.html
>>
>> Does it still happen with 0.9.3?
>>
>> http://www.erights.org/download/0-9-3/
>
> I installed 0.9.3d. With just a few executions, I didn't see the
> "OldFarRef may only be smashed..." (I debug with println and a shell.)
>
> Note that I only saw it once in over 232 executions (the number
> of files in my etrace directory). It doesn't seem to be very
> repeatable. Note that the source code doesn't have any eventual
> sends (no<- operators). I'm using E because it seems safer that
> perl and I don't know python. :-)

E uses several communicating vats even if you never create new vats in 
your own code: the start vat, a parser vat, and a eval vat (see 
eLauncherAuthor.emaker). This is why 0.9.2 often fails to start. Can we 
get 0.9.3 linked from the erights.org downloads page?

> Note that I am getting zero length files in my etrace directory.

By the way, if you're using traceln() for debugging then 0.9.3 uses 
java.util.logging, so messages will go to the console by default rather 
than to the E log.


-- 
Dr Thomas Leonard
IT Innovation Centre
2 Venture Road
Southampton
Hampshire SO16 7NP

Tel: +44 0 23 8076 0834
Fax: +44 0 23 8076 0833
mailto:tal-v5nx5w6akNyLE8xUarVfuPLx9OUvmyODWmv/[email protected]
http://www.it-innovation.soton.ac.uk