On Nov 6, 2015, at 1:01 AM, jude <flexcapaci...@gmail.com> wrote: > How did the TLF engineers figure this out?
I think you just need to know and know people who know. Microsoft’s site has some great info, but it’s scattered all over the place. I learned a lot some years ago when I was working on creating some fonts. I spent a lot of time with VOLT and Microsoft’s OpenType pages. That gave me a pretty solid background on OpenType features. Playing with AFTKO added to that. Taking apart a Devanagari font was very instructive as well. I’m well versed in Hebrew and have enough exposure to Arabic to be familiar with those issues. I have a colleague in Japan who is my go-to for Japanese issues as well as some people I know in the US who are Japanese experts. > Things like RTL, LTR, hebrew, > japanese vertical. What tests can you run to verify that what you do won't > break it? Mustella? FlexUnit. If you run ant all, it compiles TLF and runs the tests. There’s over a thousand tests in TLF which check for various things from cursor movement, to selections, to content, etc. There are some holes in the tests (i.e. we have not yet added tests for tables), but it’s pretty comprehensive. It does not have pixel comparisons, but I’m not sure you need it in most cases. One thing it should be testing for, but doesn’t is performance. That would have caught the performance bug we currently have before it was committed. I have not yet managed to tack down that bug which seems to have been introduced some time before Adobe donated the TLF code.
