[JGloss] Re: Invalid Dictionary Files?

[Rolf L. <no...@no...>]

Roy wrote:

> I had been able to use JGloss 1.0.8 and 2.0.a2 in the past without any
> problems.  I recently downloaded 1.0.10 and 2.0.a3 and unziped the files
> into their own directories.  However, now when I try to run JGloss (either
> version), I get an error that states that the dictionary files I'm using are
> invalid.  The error message when I click the "Why" button is:

>     The file C:\Program Files\Japanese\edict was tested against each
> built-in dictionary format.
>     The following is a list of all formats and the reason the file was not
> recognized:
> 
>     WadokuJT:The character encoding of the dictionary file was detected as
> EUC-JP, which is not the expected encoding UTF-8.
>     EDICT (Unicode):The character encoding of the dictionary file was
> detected as EUC-JP, which is not the expected encoding UTF-8.

> I tried deleting the index files, I tried downloading new dictionary files
> from Jim Breen's site without any success.  The weird thing is that now the
> old versions of JGloss no longer work.

Hi Roy,

it is unfortunate to hear that you are having problems.  I had similar 
trouble in the past and Michael fixed the problems quickly.  From his 
lack of responsiveness I assume that either he is busy at work at the 
moment or (more likely) away on his much deserved holidays.

I do not really have a clue as to the inner workings of the code.  I 
used version 1.0.9 up to now but upon seeing you ran into trouble I 
untarred v1.0.10 and tried it out on my Win98 machine.  Everything 
worked just fine.

Do you know what md5sums are?  Here are the check sums for the files I 
use.  You can probably get a Windows executable of md5sum.exe somewhere 
on the net and check to see whether you and I have the same version of 
the dict files.

9e7630278e6c7d76f7c3bee8a7635fe1 *Copy of edict.txt
b8fb96784446e0973c1e861fbc3fe4de *Copy of edict.txt.jjdx
fee3b20bff4b26f49dc701156a2b7bd1 *WdJTUTF.txt
165194f80ebc9371c0836598c9eb61be *WdJTUTF.txt.jjdx
f8120d29072b5d8c5b4ecf6cec84d621 *edict

Here are the sizes in bytes of the individual files.

   6212379 Copy of edict.txt
   2625744 Copy of edict.txt.jjdx
  22151547 WdJTUTF.txt
   8576024 WdJTUTF.txt.jjdx
   5272295 edict

My recollection is cloudy but I think I remember that I needed to work a 
bit on the stock edict files before 1.0.9 was ready to accept them. 
That is why I have "Copy of edict.txt".  Plain edict is there in the 
directory as well but not used at all.  IIRC what I did is rename edict 
to edict.txt, opened it in Word or OpenOffice as "Encoded text" and 
saved again as "Encoded text" but IIRC chose UTF-8 as encoding.  You 
might of course try other venues to change to encoding from EUC-JP or 
whatever it is now to UTF-8.  The resulting file was the "Copy of 
edict.txt" you see above.  I added this one successfully through the 
menu and got an index file created for it after which I used it ever 
since without a hitch.  I informed Michael about this but IIRC he was 
not able to reproduce this behaviour and we attributed it to some W98 
weirdness.

The error message you saw indicates this might be a way for you to get 
things going again.  Let us know if this resolved the problems you 
encountered.

Best

Rolf


PS: Posted & mailed.  Please reply to the list so that everyone benefits.