mirror of
https://git.postgresql.org/git/postgresql.git
synced 2024-09-07 07:19:21 +02:00
The IANA tzcode library has a feature to read a time zone file named "posixrules" and apply the daylight-savings transition dates and times therein, when it is given a POSIX-style time zone specification that lacks an explicit transition rule. However, there's a problem with that code: it doesn't work for dates past the Y2038 time_t rollover. (Effectively, all times beyond that point are treated as standard time.) The IANA crew regard this feature as legacy, so their plan is to remove it not fix it. The time frame in which that will happen is unclear, but presumably it'll happen well before 2038. Moreover, effective with the next IANA data update (probably this fall), the recommended default will be to not install a "posixrules" file in the first place. The time frame in which tzdata packagers might adopt that suggestion is likewise unclear, but at least some platforms will probably do it in the next year or so. While we could ignore that recommendation so far as PG-supplied tzdata trees are concerned, builds using --with-system-tzdata will be subject to whatever the platform's tzdata packager decides to do. Thus, whether or not we do anything, some increasing fraction of Postgres users will be exposed to the behavior observed when there is no "posixrules" file; and if we do nothing, we'll have essentially no control over the timing of that change. The best thing to do to ameliorate the uncertainty seems to be to proactively remove the posixrules-reading feature. If we do that in a scheduled release then at least we can release-note the behavioral change, rather than having users be surprised by it after a routine tzdata update. The change in question is fairly minor anyway: to be affected, you have to be using a POSIX-style timezone spec, it has to not have an explicit rule, and it has to not be one of the four traditional continental-USA zone names (EST5EDT, CST6CDT, MST7MDT, or PST8PDT), as those are special-cased. Since the default "posixrules" file provides USA DST rules, the number of people who are likely to find such a zone spec useful is probably quite small. Moreover, the fallback behavior with no explicit rule and no "posixrules" file is to apply current USA rules, so the only thing that really breaks is the DST transitions in years before 2007 (and you get the countervailing fix that transitions after 2038 will be applied). Now, some installations might have replaced the "posixrules" file, allowing e.g. EU rules to be applied to a POSIX-style timezone spec. That won't work anymore. But it's not exactly clear why this solution would be preferable to using a regular named zone. In any case, given the Y2038 issue, we need to be pushing users to stop depending on this. Back-patch into v13; it hasn't been released yet, so it seems OK to change its behavior. (Personally I think we ought to back-patch further, but I've been outvoted.) Discussion: https://postgr.es/m/1390.1562258309@sss.pgh.pa.us Discussion: https://postgr.es/m/20200621211855.6211-1-eggert@cs.ucla.edu |
||
---|---|---|
.. | ||
data | ||
tznames | ||
.gitignore | ||
known_abbrevs.txt | ||
localtime.c | ||
Makefile | ||
pgtz.c | ||
pgtz.h | ||
private.h | ||
README | ||
strftime.c | ||
tzfile.h | ||
zic.c |
src/timezone/README This is a PostgreSQL adapted version of the IANA timezone library from https://www.iana.org/time-zones The latest version of the timezone data and library source code is available right from that page. It's best to get the merged file tzdb-NNNNX.tar.lz, since the other archive formats omit tzdata.zi. Historical versions, as well as release announcements, can be found elsewhere on the site. Since time zone rules change frequently in some parts of the world, we should endeavor to update the data files before each PostgreSQL release. The code need not be updated as often, but we must track changes that might affect interpretation of the data files. Time Zone data ============== We distribute the time zone source data as-is under src/timezone/data/. Currently, we distribute just the abbreviated single-file format "tzdata.zi", to reduce the size of our tarballs as well as churn in our git repo. Feeding that file to zic produces the same compiled output as feeding the bulkier individual data files would do. While data/tzdata.zi can just be duplicated when updating, manual effort is needed to update the time zone abbreviation lists under tznames/. These need to be changed whenever new abbreviations are invented or the UTC offset associated with an existing abbreviation changes. To detect if this has happened, after installing new files under data/ do make abbrevs.txt which will produce a file showing all abbreviations that are in current use according to the data/ files. Compare this to known_abbrevs.txt, which is the list that existed last time the tznames/ files were updated. Update tznames/ as seems appropriate, then replace known_abbrevs.txt in the same commit. Usually, if a known abbreviation has changed meaning, the appropriate fix is to make it refer to a long-form zone name instead of a fixed GMT offset. The core regression test suite does some simple validation of the zone data and abbreviations data (notably by checking that the pg_timezone_names and pg_timezone_abbrevs views don't throw errors). It's worth running it as a cross-check on proposed updates. When there has been a new release of Windows (probably including Service Packs), the list of matching timezones need to be updated. Run the script in src/tools/win32tzlist.pl on a Windows machine running this new release and apply any new timezones that it detects. Never remove any mappings in case they are removed in Windows, since we still need to match properly on the old version. Time Zone code ============== The code in this directory is currently synced with tzcode release 2020a. There are many cosmetic (and not so cosmetic) differences from the original tzcode library, but diffs in the upstream version should usually be propagated to our version. Here are some notes about that. For the most part we want to use the upstream code as-is, but there are several considerations preventing an exact match: * For readability/maintainability we reformat the code to match our own conventions; this includes pgindent'ing it and getting rid of upstream's overuse of "register" declarations. (It used to include conversion of old-style function declarations to C89 style, but thank goodness they fixed that.) * We need the code to follow Postgres' portability conventions; this includes relying on configure's results rather than hand-hacked #defines (see private.h in particular). * Similarly, avoid relying on <stdint.h> features that may not exist on old systems. In particular this means using Postgres' definitions of the int32 and int64 typedefs, not int_fast32_t/int_fast64_t. Likewise we use PG_INT32_MIN/MAX not INT32_MIN/MAX. (Once we desupport all PG versions that don't require C99, it'd be practical to rely on <stdint.h> and remove this set of diffs; but that day is not yet.) * Since Postgres is typically built on a system that has its own copy of the <time.h> functions, we must avoid conflicting with those. This mandates renaming typedef time_t to pg_time_t, and similarly for most other exposed names. * zic.c's typedef "lineno" is renamed to "lineno_t", because having "lineno" in our typedefs list would cause unfortunate pgindent behavior in some other files where we have variables named that. * We have exposed the tzload() and tzparse() internal functions, and slightly modified the API of the former, in part because it now relies on our own pg_open_tzfile() rather than opening files for itself. * tzparse() is adjusted to never try to load the TZDEFRULES zone. * There's a fair amount of code we don't need and have removed, including all the nonstandard optional APIs. We have also added a few functions of our own at the bottom of localtime.c. * In zic.c, we have added support for a -P (print_abbrevs) switch, which is used to create the "abbrevs.txt" summary of currently-in-use zone abbreviations that was described above. The most convenient way to compare a new tzcode release to our code is to first run the tzcode source files through a sed filter like this: sed -r \ -e 's/^([ \t]*)\*\*([ \t])/\1 *\2/' \ -e 's/^([ \t]*)\*\*$/\1 */' \ -e 's|^\*/| */|' \ -e 's/\bregister[ \t]//g' \ -e 's/\bATTRIBUTE_PURE[ \t]//g' \ -e 's/int_fast32_t/int32/g' \ -e 's/int_fast64_t/int64/g' \ -e 's/intmax_t/int64/g' \ -e 's/INT32_MIN/PG_INT32_MIN/g' \ -e 's/INT32_MAX/PG_INT32_MAX/g' \ -e 's/INTMAX_MIN/PG_INT64_MIN/g' \ -e 's/INTMAX_MAX/PG_INT64_MAX/g' \ -e 's/struct[ \t]+tm\b/struct pg_tm/g' \ -e 's/\btime_t\b/pg_time_t/g' \ -e 's/lineno/lineno_t/g' \ and then run them through pgindent. (The first three sed patterns deal with conversion of their block comment style to something pgindent won't make a hash of; the remainder address other points noted above.) After that, the files can be diff'd directly against our corresponding files. Also, it's typically helpful to diff against the previous tzcode release (after processing that the same way), and then try to apply the diff to our files. This will take care of most of the changes mechanically.