[sword-devel] utf lexicons + bug
Tue, 25 Jun 2002 05:12:22 +0800
Content-Type: text/plain; charset=us-ascii; format=flowed
>Entries appear repeatedly. Can you confirm this? I shall attach a debug
I had similar problems with the windows version of the Sword
project. Also, especially in the area of the number entries you
mentioned ("10 MILLION" etc.), there are weird jumps when you try to
scroll using the cursor keys -- after scrolling forward n keys, the
dictionary will just jump back to a certain entry.
As I mentioned, I am also having problems with imp2ld. I managed to
compile a cygwin version today (the compress functions seem to be
missing in MingW) from the current release, so I feel confident
reporting the bugs. One of them is related to your problem.
Please find enclosed two modules, abc_imp2ld and abc_addld. abc_imp2ld
was created by imp2ld from
---- abc.imp ----
The first letter of the latin alphabet.
Being the second letter of the latin alphabet, it can be found just after a, but before c.
This is the third letter of the latin alphabet and the end of this silly dictionary.
---- abc.imp ----
abc_addld was created by calling addld manually (same entries). As you
can see, this dictionary is only a minor improvement over the "home town
The generated .dat files are identical. The generated .idx files,
however, differ. Here's the problems I encountered:
1. imp2ld crashes when given the name of a nonexistant input
file. Talk about "nice error messages"...
2. When displaying abc_imp2ld in sword, sword (windows) displays a lot
of "A" entries (no B or C). Accessing any of them will result in a
write (!) access violation. sword survives, however.
3. When displaying abc_addld in sword, sword displays the entries "A A
B C" (!). Accessing the entries works, though, but clicking on the
first A entry will cause sword to jump to the second one.
Interestingly enough, in the CEDICT the A entry will appear twice
as well. (only checked up to beta-G. Up to beta-G this was the first
entry. beta-H changed the display ordering).
4. mod2zmod can't handle abc_imp2ld as well -- endless loop. More
about that later when I have more time to test. (This was the
mod2zmod from the alpha website, my mod2zmod cygwin-compiled
executable performs only endless loops of doing nothing).
Sorry for distributing the attachment to everybody, but I felt it's
not too much of a burden, since it's only 1.5K.
My guess is that the index created by imp2ld is corrupt, and that
there is a bug in the sword library lexdict handling. Can't
investigate further right now, though.
Greetings and blessings,
firstname.lastname@example.org - http://www.web42.com/crenz/ - http://www.web42.com/
"It's so much easier to pray for a bore than to go and see one." -- C.S. Lewis
Content-Disposition: attachment; filename="abc.zip"