Added indian tables by sreeja.param@gmail.com

[egli]

These tables need some improvements before they can go in:

• Integration with the Makefile.am (I can take care of that)
• add an entry to the NEWS file about the changes/improvements
• proper copyright statements and license headers
• update of the AUTHORS file with all the authors
• when changing existing tables then retain the existing copyright notices and license headers (unless your table is a complete rewrite.
• Fix make check
• Delete the old tables

[egli]

OK, the copyright and the license should be according to this post in the mailing list.

[egli]

I think I retained all the existing copyrights when integrating the new tables

[egli]

make check doesn't pass. The problem is that both devanagari.cti and bengali.cti define the class halal. Now if you include both devanagari.cti and bengali.cti which is done in new.ctb and sat.ctb, then you get a double definition of the character class halal. Liblouis consequently barfs with ../tables/bengali.cti:191: error: character class already defined.

[egli]

So sreeja says that his tables supersede all the tables originally provided by Samuel Thibault. That would mean all of

• tables/as.ctb
• tables/awa.ctb
• tables/bengali.cti
• tables/bh.ctb
• tables/bn.ctb
• tables/bra.ctb
• tables/devanagari.cti
• tables/dra.ctb
• tables/gon.ctb
• tables/gu.ctb
• tables/gujarati.cti
• tables/gurmukhi.cti
• tables/hi.ctb
• tables/kannada.cti
• tables/kha.ctb
• tables/kn.ctb
• tables/kok.ctb
• tables/kru.ctb
• tables/malayalam.cti
• tables/ml.ctb
• tables/mni.ctb
• tables/mr.ctb
• tables/mun.ctb
• tables/mwr.ctb
• tables/ne.ctb
• tables/new.ctb
• tables/or.ctb
• tables/oriya.cti
• tables/pa.ctb
• tables/pi.ctb
• tables/sa.ctb
• tables/sat.ctb
• tables/sd.ctb
• tables/ta.ctb
• tables/tamil.cti
• tables/te.ctb
• tables/telugu.cti

Need to confirm with sreeja

[egli]

I say:

It has not been merged with master. My problem is that I don't know what to do with the old tables. There might be existing users out there that depend on them. So I'm reluctant to remove them. But if I don't remove them (or at least two of them) your tables will not work. So I'm in a bit of a pickle, you see.
Maybe we should move the tables into a subdirectory named"deprecated" and keep them for a couple of releases. We should try to stay backwards compatible.

Mesar says:

I am not aware that anyone is using them, and the point of these updated tables is that they are hopefully better, because they have had user testing. I will be happy to drop the old tables assuming we get harness tests for each touched table.

having them in a subdirectory will only complicate things i.e. people will expect things to work, but the paths will not be picked up. If we drop them, people or we can just download a old version of
liblouis, or go back in the git history.
In any case, having tests will ensure that the tables only improve in quality.

[egli]

OK, so what I'll do is to simply delete the old tables and replace them with the new ones

[egli]

OK, got rid of 15 of the old tables. Now I still have the following to go:

Table	Language
awa.ctb	Awadhi
bh.ctb	Bihari
bra.ctb	Braj
dra.ctb	Dravidian
gon.ctb	Gondi
kha.ctb	Khasi
kok.ctb	Konkani
kru.ctb	Kurukh
mun.ctb	Munda
mwr.ctb	Marwari
new.ctb	Old Newari
pi.ctb	Pali
sat.ctb	Santali
ta.ctb	Tamil
tamil.cti	Tamil

Is there a replacement for these tables, are they obsolete or should I just leave them?

[egli]

OK, thanks to sreeja we are three tables down. Now we have left the following:

Table	Language
bh.ctb	Bihari
dra.ctb	Dravidian
gon.ctb	Gondi
kok.ctb	Konkani
kru.ctb	Kurukh
mun.ctb	Munda
mwr.ctb	Marwari
new.ctb	Old Newari
pi.ctb	Pali
sat.ctb	Santali
ta.ctb	Tamil
tamil.cti	Tamil

[egli]

sreeja says:

I am attaching the files that i can give as an replacement for the list that you have given. Rest you can keep it as it will not disturb our files.

i am attaching files for awadhi, khasi and braj language that you can replace with samuel files. For tamil, ta.ctb and tamil.cti is already there in my package.

[egli]

After talking to Sreeja and Dipendra we decided to remove the tables that "cause" the conflict with Halant, namely new.ctb and sat.ctb. According to Dipendra they are not in use much.

[egli]

Now the tests pass. We still have the following old tables. We'll just leave them for now

Table	Language
bh.ctb	Bihari
dra.ctb	Dravidian
gon.ctb	Gondi
kok.ctb	Konkani
kru.ctb	Kurukh
mun.ctb	Munda
mwr.ctb	Marwari
pi.ctb	Pali
ta.ctb	Tamil
tamil.cti	Tamil

[mhameed]

Hi Christian,

The tamil tables, we need to be careful that it doesn't break any
existing functionality.
The tamil tables were worked through with the NVDA tamil translator and
a school in the region.

I don't think we have a lot of tamil tests at the moment, but hopefully
this can be addressed in an upcoming release.
Never the less, please check that they dont break the existing tamil tests.

thanks,
Mesar

[egli]

Oh, I just merged it right now. Let me check.

[egli]

@mhameed I didn't even realize that we had a test harness for Tamil

[egli]

@mhameed
Ran 4690 tests (15.37% success), with 3969 failures and 0 errors
This just sucks
Let me see how it was before

[egli]

Ah, before we had a much better result
Ran 4690 tests (99.87% success), with 6 failures and 0 errors.

[mhameed]

On Mon 01/09/14,02:23, Christian Egli wrote:

@mhameed
Ran 4690 tests (15.37% success), with 3969 failures and 0 errors
This just sucks

yes, thats why we should be careful.
So please revert anything to do with tamil.
We should not have one big pull request, but many smaller once, one per
language.
It would be more managable.

Mesar

---

Reply to this email directly or view it on GitHub:
#3 (comment)