Closed GoogleCodeExporter closed 8 years ago
Original comment by thasso.g...@gmail.com
on 6 Jul 2011 at 10:31
Can you please try with the new RC2. I was not able to reproduce the problem
right now, but if it still exists, I will take a deeper look.
Original comment by thasso.g...@gmail.com
on 6 Jul 2011 at 6:08
With RC2 the library construction runs by itself.
However, there are two errors during the process, second one being "fatal". See
attached log.
I'm not sure what are any of those error statements trying to say.
Original comment by peter.gl...@gmail.com
on 7 Jul 2011 at 3:04
Attachments:
Okey, one minor bug that I just fixed is that actually the first error was
already fatal.
What happens is that we try to read column 6 (field 5 in the error message
because we start counting at 0) from the .pro profile file. For some reason you
get an error that the column does not exist. Its the last column of a line (at
least after the initial profile run) in the .pro file and contains the absolute
number of molecules. for example:
chr1:11874-14409W uc001aaa.3 NC 1652 4.3999683202280946E-5
220
here, the 220 at the end is what we are looking for.
Can you attache your .pro file, I guess its f0.pro. If that is in fact invalid
(which I am assuming), it would be great if you could also attach the .par
parameters file and a compressed version of the .gtf annotation file. Otherwise
I need you .pro to check for any strange characters. For that matter, are you
on Windows or on a Unix box ?
Original comment by thasso.g...@gmail.com
on 7 Jul 2011 at 4:42
See attached archive.
You were right with the .pro file. The last line is not right.
The system is CentOS 5, 64bit (I think it's Oracle HotSpot JRE). However, it's
a new server so I tried it on the older one with Ubuntu Serer 64bit (OpenJDK),
and it works OK. So I wasn't able to reproduce the error on the second machine.
But, in case that you are still interested ;-) I tried generating the .pro file
a couple times (on the new server) and the last line gets corrupted, it seems
like the the output wasn't "flushed" before the file was closed. Examples:
chr1:230838274-230850336C uc009xfe.2 CDS 1929
2.2080220802208022E-4 1104
chr1:230838274-230850336C uc009xff.2 CDS 2495 1.6<no end of
line>
chr1:226548393-226595801C uc001hqd.3 CDS 3984
7.87996848012608E-5 394
chr1:226723319-226730469C <ends here>
chr1:235824345-236046940C uc001hxi.2 CDS 8291
5.9600238400953606E-5 298
chr1:235824345-236046940C uc001hxj.2 CDS 13476 1.056004224
Original comment by peter.gl...@gmail.com
on 7 Jul 2011 at 5:29
Attachments:
Sorry, forgot to mention, it's not just the last line (which can be judged from
different "last" transcripts) the number of lines changes as well.
So it obviously just doesn't finish writing.
Original comment by peter.gl...@gmail.com
on 7 Jul 2011 at 5:33
Thanks for the info. I will check for any missing close() or flush()
statements.
Just out of curiosity, can you check which Java version is used on your CentOS
machine ( the one where the error occurs ). Maybe I could install the same
version here and reproduce the bug.
Original comment by thasso.g...@gmail.com
on 7 Jul 2011 at 5:46
The java version is:
$ java -version
java version "1.6.0_26"
Java(TM) SE Runtime Environment (build 1.6.0_26-b03)
Java HotSpot(TM) 64-Bit Server VM (build 20.1-b02, mixed mode)
We have multiple installations, so I also tried:
java version "1.6.0_20"
OpenJDK Runtime Environment (IcedTea6 1.9.8) (rhel-1.22.1.9.8.el5_6-x86_64)
OpenJDK 64-Bit Server VM (build 19.0-b09, mixed mode)
And the problem persists.
Original comment by peter.gl...@gmail.com
on 7 Jul 2011 at 6:06
Hi,
as Thasso already suggested
(http://code.google.com/p/fluxcapacitor/issues/detail?id=52#c1), for enhancing
compliance with the Flux library, or Java code in general, I would recommend to
prefer the Oracle (Sun) JDK over the OpenJDK. We had already heavy issues due
to different thread handling, and if I read that report right, it seems that
file I/O also differs. See also the opinion posted there:
http://stackoverflow.com/questions/4422808/oracle-sun-jdk-vs-openjdk
My 5 cents,
micha
Original comment by gmicha@gmail.com
on 7 Jul 2011 at 6:48
[deleted comment]
I'm experiencing the same problem here with FluxSimulator-1.0-RC2 on CentOS 5.6
and this java version:
java version "1.6.0_24"
Java(TM) SE Runtime Environment (build 1.6.0_24-b07)
Java HotSpot(TM) Server VM (build 19.1-b02, mixed mode)
I'm running the demo dataset without modifying anything. Same exception as
Peter (however the index causing the ArrayIndexOutOfBoundsException can
change). As for Peter, the output of the .pro file is not flushed (it has a
"perfect" size of 409600 bytes!).
Original comment by mikael.s...@gmail.com
on 12 Jul 2011 at 7:24
I found the flush problem and its fixed. The fix will be incorporated in the
next release (RC3).
Original comment by thasso.g...@gmail.com
on 12 Jul 2011 at 11:23
Original issue reported on code.google.com by
peter.gl...@gmail.com
on 5 Jul 2011 at 7:48