FreeBSD port lang/squeak: 4.4.7_2 - invalid UTF8 - UTF8TextConverter class>> errorMalFormedInput:

Hi, I'm a maintainer of FreeBSD port lang/squeak.

I receive a serious problem report below:
- http://www.freebsd.org/cgi/query-pr.cgi?pr=ports/171291

In this problem report, only open browser with Form class, raise debugger "invalid UTF8 - UTF8TextConverter class>> errorMalFormedInput:".
Full SqueakDebug.log is here:
- http://qml.610t.org/squeak/mutoh_20120904.files/SqueakDebug.log

I can repeat this problem with FreeBSD port VM.

Some other VMs don't have this problem below.
- Version 4.3 at MacOS X 10.8.1
-- http://ftp.squeak.org/4.3/Squeak-4.3-All-in-One.zip
- 4.4.7-2357 VM at Debian Linux
- 4.4.7-2357 VM at FreeBSD http://squeakvm.org/
-- http://squeakvm.org/unix/release/Squeak-4.4.7.2375-freebsd8.2_i386.tar.gz

There is this problem at FreeBSD port VM, but not squeakvm.org VM.
So I try to find difference both VMs.

FreeBSD port VM build log is here:
- http://qml.610t.org/squeak/mutoh_20120904.files/squeak.20120906.txt

If there are  some idea to fix this problem, any information is welcome.

Thanks in advance.
--
Takeshi MUTOH <[hidden email]>

 
On 06.09.2012 08:06, Takeshi MUTOH wrote: That's seems to be a corrupted Squeak41.sources file, afaict the source
offset in the image is correct, but the contents at that offset is
gibberish.
I'd suggest checking the version included in FreeBSD port vs the one
found at ftp://squeak.org/4.1/SqueakV41.sources.gz


Cheers,
Henry

Hi Henry, thank you for your advice.

On 2012/09/06, at 18:11, Henrik Sperre Johansen <[hidden email]> wrote:

> On 06.09.2012 08:06, Takeshi MUTOH wrote:
>>  Hi, I'm a maintainer of FreeBSD port lang/squeak.
>>
>> I receive a serious problem report below:
>> - http://www.freebsd.org/cgi/query-pr.cgi?pr=ports/171291
  (snip)
> That's seems to be a corrupted Squeak41.sources file, afaict the source offset in the image is correct, but the contents at that offset is gibberish.
> I'd suggest checking the version included in FreeBSD port vs the one found at ftp://squeak.org/4.1/SqueakV41.sources.gz

Your suggestion is right.
When I replace Squeak41.sources to ftp://squeak.org/4.1/SqueakV41.sources.gz, the problem is fixed.

I'll make patch to fix this problem.

How do you find the problem?
I want to know for fixing this kind of problem.

Thank you very much, again.
--
Takeshi MUTOH <[hidden email]>

 

 
Hi Henry, thank you for your advice.

On 2012/09/06, at 18:11, Henrik Sperre Johansen [hidden email] wrote:

On 06.09.2012 08:06, Takeshi MUTOH wrote:

 Hi, I'm a maintainer of FreeBSD port lang/squeak.

I receive a serious problem report below:
- http://www.freebsd.org/cgi/query-pr.cgi?pr=ports/171291

  (snip)

That's seems to be a corrupted Squeak41.sources file, afaict the source offset in the image is correct, but the contents at that offset is gibberish.
I'd suggest checking the version included in FreeBSD port vs the one found at ftp://squeak.org/4.1/SqueakV41.sources.gz

Your suggestion is right.
When I replace Squeak41.sources to ftp://squeak.org/4.1/SqueakV41.sources.gz, the problem is fixed.

I'll make patch to fix this problem.

How do you find the problem?
I want to know for fixing this kind of problem.

Thank you very much, again.
--
Takeshi MUTOH [hidden email]

From the stacktrace and error message.
Error says something can't be read as UTF8 (which you describe as happening when clicking on the class), a bit down in the trace it is clear that something originated as a class comment:
ClassOrganizer(BasicClassOrganizer)>>classComment

In one of the top frames:
UTF8TextConverter class>>decodeByteString:
    Receiver: UTF8TextConverter
    Arguments and temporary variables:
        aByteString:     '*garbled contents not remotely resembling UTF8*

Yeah, that's something which should not be decoded as UTF8 in the first place...
a bit further down:

RemoteString>>text
    Receiver: a RemoteString
    Arguments and temporary variables:
        theFile:     MultiByteFileStream: '/usr/local/lib/squeak/4.4.7-2383/SqueakV41.sourc...etc...
    Receiver's instance variables:
        sourceFileNumber:     1
        filePositionHi:     4185358

file name/location looked ok, filePositionHi checked against a 4.1 image where the error did not occur was the same.
Hence, it seemed likely problem was a corrupt .sources file (which the RemoteString points to) rather than the method in .image file containing an incorrect source offset.

Cheers,
Henry