[PATCH v9] Locales: Cyrillic -> ASCII transliteration table [BZ #2872]

[Rafal Luzynski]

10.12.2018 22:20 Marko Myllynen <myllynen@redhat.com> wrote:
> 
> Hi,
> 
> On 08/12/2018 03.15, Rafal Luzynski wrote:
> > [...]
> > Marko, it would be good to hear your opinion about System A vs. System B
> > again.
> 
> I think System A is a better option as it should be the same as ISO 9
> and perhaps also produces results in some cases which are more expected
> than with System B (if the Wikipedia ISO 9 article is to be believed).
> 
> Wrt BZ #2872 I think it's good to keep it in mind but IMHO we can also
> deviate from it if needed, however with System A + ASCII fallback
> definitions the RFE should be satisfied as well?

That's exactly what I meant (sorry if it was not clear before).

> > [...]  Marko, what is your
> > opinion about possible implementation of SFS 4900 in these cases:
> > 
> > * When the destination charset does not contain required Latin diacritic
> >   characters (e.g., it is plain ASCII)?
> 
> This would be according to http://jkorpela.fi/iso9.html8 so for example
> instead of ž -> zh and instead of štš -> shtsh.

Agree.

> > * When the output is ambiguous, that means, when two different Cyrillic
> >   strings produce the same Latin (or ASCII) output?
> 
> This is a good point and one I haven't considered but I'm not sure is
> there anything we can do about this (at least without major locale
> system internals work)?

I agree with the suggestion that we can't do much about it.  I mean,
there are possibly solutions (like using more punctuation characters)
but they don't look natural to me.

> Do you have any rough idea how frequently this
> could happen or is this more a theoretical issue? (Sorry if I've missed
> earlier comments about this, it's been a long thread.)

Yes, Egor provided this example many times:

"схема" -> "shema" (if "с" -> "s" and "х" -> "h")
"шема"  -> "shema" (if "ш" -> "sh")

I don't think that it matters how frequent are these cases.  I think that
the question is if ambiguity is a bug because if yes then even one corner
case proves that the solution is wrong.

> [...]
> Yes, I also think System A AKA ISO 9 would be a better choice but I'll
> leave the final decision for you two (and others who might weigh in).

Egor is a native speaker so I respect his opinion even if I'm not fully
convinced for technical reasons.  Sadly, nobody else provides any opinion
which could weigh.  I am going to write a separate email about it.

Regards,

Rafal