problem with qif-parse.scm in trunk

[David Reiser]

On Sep 17, 2008, at 9:00 AM, Derek Atkins wrote:

> David Reiser <dbreiser at earthlink.net> writes:
>
>>> Looks like it is coming from Apple's libSystem.dylib. I'll see if I
>>> can find someone who knows about any Apple regex oddities.
>>
>> Well, here's some Apple regex oddity: The regex parser apparently
>> doesn't like higher-than-ascii utf-8 unless the file it's working on
>> starts with a utf-8 BOM (0xefbbbf). I can't even grep qif-parse.scm
>> for the GBP symbol -- there is no output from the grep command. If I
>> use bbedit to prepend the three hex bytes to the file, then grep
>> successfully finds the symbols. But if the .scm file starts with the
>> BOM, gnucash launch now fails with:
>
> Interesting!  Do you have the Gnu Regex library available?  What if
> you explicitly link against that instead of using the regex in Apple's
> libc?
>
> -derek

I don't know how hard it will be to get gnu libc. I'll keep working on  
that.

In the meantime, I've discovered that the BOM isn't the issue. I dug  
out a hex editor and discovered that bbedit's save as utf-8 (with no  
BOM) converts all the GBP symbols from hex A3 to hex C2A3. Using that  
encoding on the mac (with no BOM at the file beginning) allows gnucash  
trunk to run. No other characters seem to be affected. Since saving as  
utf-16 converts A3 to 00A3, I'm confused by the encoding shenanigans.

Dave
--
David Reiser
dbreiser at earthlink.net