World: r4wp

test-failures 55? Are those the rounding issues?

Yes.

Why don't I experience the problem?

Older GCC, maybe?

Let me try building with an older GCC to see if that changes anything.

how do I check the gcc version?

gcc -v

gcc version 4.6.3 (Ubuntu/Linaro 4.6.3-1ubuntu5)

...and glibc version?

Look in the package manager

apt-cache show libc6 | grep Version

[ladislav-:-lkub64]:/r3/make$ apt-cache show libc6 | grep Version
Version: 2.15-0ubuntu10.3
Version: 2.15-0ubuntu10.2
Version: 2.15-0ubuntu10

Andreas, it looks like you have a new test as well?

(Total 4344)

I found the "total" number to be not particularly reliable.

GCC 4.3.2 compiled binary doesn't have the round regressions:

16-Dec-2012/23:42:06+1:00 2.101.0.4.2 Total: 4333 Succeeded: 3950 
Test failures: 53 Crashes: 11 Dialect failures: 0 Skipped: 330

(Differing total again.)

Hmm, that may need a fix...

The last result is GCC 4.3.2 + glibc 2.7 at compile time, glibc 2.16 
at runtime.

How come you have Crashes: 11?

How would I find out?

run-safe and diff the full log?

To see what's missing?

those are the ones marked as #crash

But I have "crashes: 11" for RT-built A111 on this machine as well.

#r3crash

yes, that is in the test source. I guess that you do not use the 
latest tests

I use the tests as cloned a few hours ago.

aha, let me check

I think the additional guess is another real regression on the machine 
I'm using. But let me find out which test is newly crashing.

No, that is not how it works

It just counts the #r3crash?

It just counts the #r3crash - 7 times and #test3crash - 3 times

(in my file)

strange

Well, then it is definitely my set of tests.

#r3crash is supposed to crash the interpreter, #test3crash is suppossed 
to crash the test framework

Ah, correct. I accidentally had an additional test copied over.

Built with GCC 4.3.2:

17-Dec-2012/0:03:50+1:00 2.101.0.4.2 Total: 4333 Succeeded: 3950 
Test failures: 53 Crashes: 10 Dialect failures: 0 Skipped: 330

Built with GCC 4.7.2:

17-Dec-2012/0:03:45+1:00 2.101.0.4.4 Total: 4333 Succeeded: 3948 
Test failures: 55 Crashes: 10 Dialect failures: 0 Skipped: 33

The round testcases.

17-Dec-2012/0:03:50+1:00 2.101.0.4.2 Total: 4333 Succeeded: 3950 
Test failures: 53 Crashes: 10 Dialect failures: 0 Skipped: 330
 - hmm, that total is not correct

Should be 4343

(the total is simply the sum of the numbers)

 Total: 
 succeeded + test-failures + crashes + dialect-failures
				+ skipped

(that line is from %test-framework.r)

Carl merged my update almost immediately

Good timing :)

What about the total, Andreas? That is suspicious...

(never experienced that here)

Is carl going to spring an R3 web chat on us? The last message in 
chat:
#8073: post from Carl 26-Nov:

Yes, the web interface is also working. Top level is a bit slow, 
but other links
 seem to respond ok.

I noticed he didn't answer in the blog when I asked about chat being 
open sourced. Could he be planning to take it commercial?