> Hi,
> 
> I was told to repost this here by Ryan Mansfield:
> 
> I just spent the last 3 days tracking down an issue which turned out to be a 
> problem with the pow function in libm  (pow(x,2.0f) was 
> returning inf. for small values of x).  I was very suprised (unhappy) at the 
> end of the struggle to find out that this is a known issue.
> 
> Downloading the new libm solved this problem.  I have a few questions
> 
> 1) how do I identify the version of the libm libraries to find out if this 
> issue is lingering elsewhere.
> 
> 2) which versions of the libm library have this issue.

I'll let someone else comment on these...
 
> 3) does QNX not have some mechanism to notify clients when something as 
> critical as the math library being broken is 
> identified.  We are building mission critical applications and I only ran into
>  this issue by fluke. I would have 
> expected something like a RECALL being issed for something this severe.

This bug was not in our system so we did not have visibility to it to take action. As part of the new 6.4 release of the
 OS, we have updated libm and this issue is corrected. We do have a defined process to notify our customers about "
critical" issues but as noted, this was not in our Problem Report system. 

> 4) do you have a list of other critical issues such as this which exist so I 
> know what else to look out for instead.  I am now uncomfortable about the 
> devices I have out in the field.

With each new release of software, we develop release notes which lists out the important updates/fixes associated with 
the release. At this time, we do not have a mechanism to make our internal bug tracking system publicly available but 
this is under discussion as part of the future Foundry 27 plans. In the interim, for projects where we are doing 
development in the open/live (Kernel project, Networking project...), you are seeing realtime/live bug fixing being done
 and posted for download in realtime.