Couldn't find end of Start Tag q_code

[sorashi]

I'm trying to run GeekJDict on Windows using Strawberry Perl, but while running the setup command:

geekjdict --update=kanji --lang=eng kanjidic2.xml.gz Unihan.zip cangjie5.txt

I ran into this problem (XML parser failing?):

Entity: line 4331: parser error : attributes construct error
<q_code qc_type="sh_des4g m_lang="fr">trame (tissage)</meaning>
                                  ^
Entity: line 4331: parser error : Couldn't find end of Start Tag q_code
<q_code qc_type="sh_des4g m_lang="fr">trame (tissage)</meaning>
                                  ^
Can't use an undefined value as a subroutine reference at C:/Users/user/AppData/Local/geekjdict/lib/GeekJDict/Update.pm line 84.
Issuing rollback() due to DESTROY without explicit disconnect() of DBD::SQLite::db handle dbname=file:C:\Users\user\.geekjdict.sqlite?mode=rwc at C:/Users/user/AppData/Local/geekjdict/lib/GeekJDict/Update.pm line 84.

This error output seems to be referring to kanjidic2.xml at line 4331

<dic_ref dr_type="sh_kk2">1342</dic_ref>

But the trame (tissage) text can be found at line 4288:

<meaning m_lang="fr">trame (tissage)</meaning>

And the q_code tag is normally closed

<q_code qc_type="sh_desc">6a10.7</q_code>

Where is the problem? Is there anything I can do to make it work?

[sorashi]

I can see GeekJDict depends on a lot of Linux internals, it might be very hard to run it on Windows...

One thing I had found out before even attempting to run GeekJDict using Strawberry Perl -- GeekJDict can't be run using WSL (Windows Subsystem for Linux) because of this issue microsoft/WSL#2395

[sorashi]

By the way -- I tried GeekJDict on my Ubuntu VPS and it's awesome. You did a very VERY good job. That's a lot of functionality. I think I will make a VM just to use GeekJDict.

If the reason for my problem doesn't come to your mind, just close this issue :)

[kroki]

I couldn't reproduce the problem with latest kanjidic2.xml.gz on Linux. Could it be the problem with XML::LibXML::Reader (my version is 2.0134) bundled with Strawberry Perl? At least for ActiveState Perl (another Perl for Windows that you may want to give a try) they tweak original Perl modules to make them work on Windows. BTW, you also have to use terminal that supports UTF-8, wide characters and 256 colors.

In any case, Windows is not officially supported, so closing with WONTFIX.

[kroki]

RE you "Windows installation" instructions: I don't know what WSL is, but GeekJDict writes to DB only when performing --update, never during normal operation (except at the start where it may have to rollback incomplete transaction if previous --update failed and DB is left in a dirty state). So if you are careful to never run --update and normal dictionary mode in parallel (and it is advised to always restart GJD after --update anyway), to my understanding you can't hit microsoft/WSL#2395