[ts-gen] mem alloc exceptions seen post 091030 [was: changes from version ... ]

[Bill Pippin]

Josef,

Thanks for the report.

> I tried to run the newest trading-shim version but it failed with
> versions newer then 091030.

I understand you to be saying that make runs fine, and the resulting
executable stops with exception messages related to memory allocation,
depending on the version, one of the following:

> ... throwing ... [lib] allocation_failure	[091102]
> ... throwing ... [lib] block_must_be_page	[091123]

All though this is not directly applicable to solving your current
problem, you may be interested to know that exceptions detected during
shim operation fall into three categories:

    1. warning exception, application code
    2. failure exception, application code
    3. library exception, component   code

The 1st class produce warning messages, and the shim continues; the 2nd
group give problem messages, and the shim exits, e.g., as for no dbms
server:

> An attempt to connect to the database ... appropriate login info.
> ...
> Problem: 511 could not initialize mysql connection
> Exiting

The 3rd kind of exception is also fatal, and is not yet caught, for no
better reason than that I'm still working on the code, and haven't yet
systematized the error cases with numbered codes as for the dbms server
example above.  The exceptions you are seeing are in this 3rd category.

I'll be putting out another release later this afternoon, and possibly
later this evening, and please consider trying these latest versions.
I have been working with the memory allocation code, and newer releases
are meant to be improvements over old ones.

That being said, neither of these exceptions should occur in any case,
and I am not seeing them, so I'll need more details.  Murphy's Law being
what it is, you may well see the same problem with the newest tarball,
but at least we'll be working from the same song sheet.

Now, as for the details, I'd like to know, for the system where you've
observed the shim to fail, the outputs produced by the following
commands:

    mysql_config --version
    g++ --version
    uname -a
    cat /proc/cpuinfo

I'd appreciate your providing the information from the above commands,
so if you'd capture their outputs and post to the list, my thanks.

> Any [other] suggestions?

The failure you are seeing is happening without a mode argument, very
early during startup, so the shim hasn't even tried to connect to the
database, or the IB tws.  The messages you are seeing suggest that
the compiled executable is completely unable to run on your system,
perhaps because I've accidently added some arch dependency to the
library code, or because there is some flaw in the way the sources
were compiled.

I don't know what the problem you're seeing is, but I'm sure we can
track it down if you are willing to help with reporting.  Please 
consider trying the newest tarballs, let me know what you find, and
understand that several rounds of debugging may be required before
we solve the problem; without a reproducing test case on a local
platform, problem causes can be elusive.

And to recap: I haven't yet pushed today's release --- I'm sending
this message now so you know what's going on, and you should see
a new tarball appear some time in the next hour, and possibly a
newer one later this evening.

Thanks,

Bill