On Wed, 27 Mar 2019 at 17:40 Markus Scherer <markus.icu@gmail.com> wrote:
Hi Tom & SG16,

First, sorry for having dropped off -- I have been swamped with other work and won't make it to today's meeting either.

Second, I would like to ask you to consider if it's necessary to add Unicode properties APIs in the language runtime.
There are widely used libraries like ICU which provide this and more.

Many users will want to be able to use the latest version of Unicode, which will tend to be newer than what their compiler provides.
There are also enough changes in Unicode properties that data structures or parsers etc. sometimes need to be adjusted, so you have a maintenance burden.
(I have been doing this for some 19 years.)

Nothing precludes an implementation to defer to ICU both at compile time and run time.
Although i found relying on ICU shipped on platforms to be problematic - ICU might keep up to of Unicode but OS definitively don't keep on top of ICU.
It's also difficult to deploy ICU on memory constraint devices or devices who can't allocate / throw exceptions / etc

And finally, I personally think that the ROI for the name property is low. As noted in the document, the data is large, but also a long \N{dozens of letters} string is not very readable. I find it's just as easy to use \uhhhh escapes with a simple code comment for which character that is, and if it's obvious (like a regular printable letter) you use the character itself anyway.

\N is a separate paper - namely https://wg21.link/p1097r2
I think there is some valid uses cases for name ( for example if you are doing an editor, ide, etc or any kind of input checking), it might be more user friendly to say "unexpected space at line 1" rather than "unexpected \U00020"

Whether that is a valid enough use-case to warrant being in the standard is up for debate - if implementers use icu, the cost of implementation is low


Best regards,

On Wed, Mar 27, 2019 at 8:42 AM Corentin <corentin.jabot@gmail.com> wrote:
As requested by Tom, please find attach D1628R0 which will be discussed during today's meeting \N{WHITE EXCLAMATION MARK ORNAMENT}

Feedback welcome :)

SG16 Unicode mailing list