[tz] [PROPOSED] Rename internal identifiers to match RFC 8536

[Paul Eggert]

On 6/17/19 8:50 AM, Brian Inglis wrote:
> Many systems base their code on the reference code provided

Sure, but the proposed change takes this into account. It does not 
change any identifier that is external or public or visible to any other 
part of a system; it affects only internal identifiers. The only names 
changed were identifiers like "tt_gmtoff" that are visible only inside 
localtime.c, identifiers like "r_todisgmt" that are visible only inside 
zic.c, and the identifier tzh_ttisgmtcnt in the file tzfile.h that is 
prominently marked as being internal to tzcode. Changes to internal 
identifiers are relatively routine and have occurred over the years 
without incident; for example, zic.c's internal identifier "puttzcod64" 
was renamed to "puttzcodepass" in 2019a.


> changing an existing de facto standard to match a proposed RFC is going the
> wrong way around
This statement doesn't match the process as I understand it. Internet 
RFC 8536 was coauthored by Arthur David Olson, Ken Murchison and myself, 
and its drafting process (which was announced here, was open to all 
commenters, and which responded to quite a few comments) did its best to 
come up with regularized names to avoid confusion that is common in this 
area, confusion that in some cases extended to tzdb itself. The recent 
changes to what is mostly commentary in tzdb attempts to take advantage 
of the RFC 8536 work by propagating some of its terminological cleanups 
back into the code.

In the long run, RFC 8536 will likely be more important than tzcode for 
parties interested in interoperability, just as Internet RFC 1952 is now 
more important than the gzip source code for parties interested in 
developing software that can read or write gzip compressed data. So 
there should be a long-term benefit to encouraging harmonization between 
internal tzdb terminology and the RFC's terminology.