[liblouis-liblouisxml] Re: Further test data for UEB
- From: Ken Perry <kperry@xxxxxxx>
- To: "liblouis-liblouisxml@xxxxxxxxxxxxx" <liblouis-liblouisxml@xxxxxxxxxxxxx>
- Date: Thu, 28 Aug 2014 17:01:08 +0000
We will have a look but I also want to mention a few of the errors in that test
you just ran might be fixed if you put in my latest pull request.
Ken
-----Original Message-----
From: liblouis-liblouisxml-bounce@xxxxxxxxxxxxx
[mailto:liblouis-liblouisxml-bounce@xxxxxxxxxxxxx] On Behalf Of Mesar Hameed
Sent: Wednesday, August 27, 2014 3:20 PM
To: liblouis-liblouisxml@xxxxxxxxxxxxx
Subject: [liblouis-liblouisxml] Re: Further test data for UEB
Hi Ken, forgot to say this in my previous email.
At the moment we have a reasonable coverage of words and contractions, but less
so for non-alpha characters.
Can you please have a look at:
http://liblouis.org/posts/minimum_set_of_tests_for_a_table/
and make a few tests, exactly as you did with the wordlist, but now with a few
symbols, transitions etc.
Again if you could get them verified by a human transcriber that would be great.
Mesar
On Wed 27/08/14,18:03, Mesar Hameed wrote:
> Hi Ken,
>
> Just to Update.
>
> Once the provided output is converted to unicode, the provided 99149
> wordlist adds 26129 "<word>'s" tests.
> adds 226 new words.
>
> The results of the UEB test harness after these additions are:
> Ran 99299 tests (98.21% success), with 1780 failures and 0 errors.
> Please note in the above line failures means braille translation
> errors, and errors means code crashes.
>
> If you want a list of these 1780 translation errors so that you can
> see if you can find suitable patches for them in the UEB table I am
> happy to provide the output if you are unable to produce it locally.
>
> Mesar
> On Wed 27/08/14,14:19, Mesar Hameed wrote:
> > Hi Ken,
> >
> > This is why most test data is stored in unicode braille, that way
> > such issues can be avoided.
> > We also do not have to worry about what codepage/dot mapping is
> > needed on the local printer.
> > Basically isolates/Compartmentalizes each area which makes it faster
> > to identify where the error is happening.
> >
> > I will be writing something that converts the brf file to unicode
> > braille before I make a harness test file out of the lot.
> >
> > thanks,
> > Mesar
> > On Wed 27/08/14,11:36, Ken Perry wrote:
> > > One warning on the files I sent. Because we do not capitalize our
> > > characters you need to capitalize our results to compare to duxbury.
> > > This includes things like brackets. I still wonder if Liblouis should
> > > not do like Duxbury and capitalize the results. I realize that would
> > > break people using the current liblouis. I find we end up converting
> > > them so we don't get the wrong dot 7 through but maybe there should be a
> > > flag in liblouis to capitalize the results. What do you think?
> > >
> > > Ken
>
>
For a description of the software, to download it and links to
project pages go to http://www.abilitiessoft.com
Other related posts:
