Conversation
Not having a unique identifier messes up learning. Complains about unknown character.
Sometimes this combination goes across word boundaries, and it is thus more useful to encode it this way.
|
I have also made some minor corrections to the existing mappings. |
This allows varnam to learn that 'kurangu' corresponds to 'குரங்கு ', and not 'குரஙு ', for instance
Previous block prevented certain common words from being typed (அந்த, இந்த, எந்த... )
44e0f67 to
4133782
Compare
|
Hi @Kishore96in, I'd like to know how Tamil is written in computers. Can you please let us know here if there are any other transliteration software ? Do you use Varnam yourself ? |
On Windows, Google input tools is available (and is actually pretty good). Apparently they also added Tamil transliteration natively in Windows 10 some time last year. I am not aware of any transliteration software other than varnam which works natively on Linux and supports Tamil (Google input tools has a chrome extension, but I don't think it can be used in other applications). Currently, I am on Linux, and so I am using varnam (specifically, the ibus interface). |
|
Thanks for the response @Kishore96in, are there any key-mapping transliteration tools like swanalekha for Tamil ? What about on the web ? Google Input Tools is the majorty used ? |
I don't know enough to comment on the usage statistics, sorry. |
|
@subins2000 This is good to merge I guess. |
|
@navaneeth Just tested this PR, scheme compiled : |
This reverts commit ccfbf5b.
Now, we can write it as 'tra', instead of the unnatural 'RRa'.
For some reason, I had made these secondary even if there were no clashes.
|
Hey @Kishore96in I have made a Go port of libvarnam. Would like to know the Tamil scheme change you made works alright in GoVarnam too: https://github.com/varnamproject/govarnam Is your changes anyway related to this problem ? varnamproject/govarnam#1 |
I have added the letter combination which is usually used for 'fa' (ஃப), and added 'za' to the list of sounds representing 'ஜ'.