[smc-discuss] Automated Rendering Testing - Status update

[Rajeesh K Nambiar]

On Mon, Sep 9, 2013 at 6:57 PM, Vasudev Kamath <kamathvasudev at gmail.com> wrote:
> On 14:18 Mon 09 Sep     , Rajeesh K Nambiar wrote:
>> In general:
>> Test cases for Meera and Rachana are reviewed and completed. The main
>> rendering testing program (rendering_test.py) is also in a good shape
>> where other interested people can start using it. We can help with
>> some tips on how to integrate rendering testing for your language,
>> send us a note or ping on #smc-project [Vasudev, I'm looking at you
>> :-) ]
>
> Well right :-). First I need to go through the code + documentation to
> see where I should start.
>
>>
>> On Mon, Sep 9, 2013 at 12:01 PM, Nandaja Varma <nandaja.varma at gmail.com> wrote:
>> > Hello,
>> >
>> > Here is the present status of the project.
>> >
>> > * The testing framework now can evaluate words with multiple correct
>> > renderings, provided the correct renderings are provided in the reference
>> > file separated by semi colon.
>> >
>> > * Reference glyph for both Rachana and Meera has been updated as per the
>> > latest updates (changes in glyph names) in the upstream.
>> >
>> > * Reference for Devanagari font is being added to the repo.
>
> Just wondering why we need to add fonts to repo. Why can't we use fonts
> from system and allow user to tell font name from conf or command line?
> I see adding font to this codebase as redundant + font might get
> outdated if upstream is active (like smc).

No, there is no need to have the fonts in repo and you have the option
to specify path to the font.

>
> Also if the tool is to be packaged then packager should strip of font
> from tool instead of that lets opt for best practise and stop
> duplicating the code/font.
>
> Disclaimer: This is written without reading or looking code, If I've misunderstood
> something please let me know.
>
>> >
>> > Present status of the framework is:
>> >
>> > * rendering_test.py can accept  up to 7 inputs, which being the test cases
>> > file, reference file, rendered output file, font file, output file, error
>> > file and a directory name.
>
> Nitpicking - A man page would be helpful ;-).

There's a ./rendering_testing.py --help :-)

>
>> > * Of this everything but reference file and rendering output are optional.
>> > * Output will be produced as per the parameters passed.
>> > * pep-8 errors reported before has been cleared.
>> >
>> > By the end of this week, I am planning to finish:
>> >
>> > * Complete Devanagari references
>> > * The immediate next priority being C++ implementation of the code, I will
>> > be working on that.
>
> Rajeesh is there any real need of C++ code? or is there any plan to
> merge this tool to harfbuzz code?

If possible, but that's a larger discussion.

>
>
>


-- 
Cheers,
Rajeesh