R devel - Jan 2023 - proposal for use ICU for timezone convertion on windows and a draft patch

Dear All,

 Last week I found an issue about R Sys.timezone function on Windows and send
email to this list. This week , I read quite a few  docs and codes about windows
time zone to IANA time zone conversion. and found in unicode.org's  ICU
library has a function  ucal_getTimeZoneIDForWindowsID which do the time zone
convert. Microsoft also use it for time zone convertion in dotnet 6 and
7.(https://github.com/dotnet/runtime/blob/main/src/native/libs/System.Globalization.Native/pal_timeZoneInfo.c)
.

IMHO, use this function to implement time zone convert on windows is better than
R's hand-write conversion implementation, use the ICU , it will update the
conversion when unicode org upgrade the ICU library and no more hand-write
conversion code needed. Also RTools already included the ICU library,and R on
windows build link ICU lib on default  , so no additional library or tools need
for use this function for time zone conversion.

so I wrote a quick and dirty patch use the ucal_getTimeZoneIDForWindowsID, I
think maybe better discuss here before I send patch to bugs.r-project.org.
I put the patch on
https://github.com/armgong/misc-r-patch/blob/main/registryTZ-ICU.diff.

this patch mainly do following three steps : 1 read current English time zone
name from registry ; 2 read current region name from registry(this also affect
conversion); 3 use ucal_getTimeZoneIDForWindowsID implement the conversion.

please feel free to share your thoughts or feedbacks of this patch.  if it
doable ,then I will modify it and send it to bugs.r-project.org

best regards,
Yu Gong




	[[alternative HTML version deleted]]

still work on it.  this morning , I found the better and short sloution of this,
the ucal_getDefaultTimeZone will return IANA formart zone name on windows.
so the main code for this issue listing  here:

/* Longest currently is 31 chars */
static char Olson[64] = "";
/* use ucal_getDefaultTimeZone to get IANA default time zone name */
const char *getTZinfo(void)
{
    UChar defaultzone[64];
    UErrorCode status_default = U_ZERO_ERROR;
    int32_t i = ucal_getDefaultTimeZone(defaultzone, ARRAYSIZE(defaultzone),
&status_default);
    if (U_SUCCESS(status_default)) {
        wcstombs(Olson, defaultzone, 64);
        Olson[63] = '\0';
    #ifdef DEBUG
        printf("TZ = %s\n", Olson);
    #endif
        return Olson;
    }
    else{
        return "unknown";
    }
}

I still maintain a longer version of this patch, but it should not needed if 
ucal_getDefaultTimeZone  is reliable .
the longer version  when  ucal_getDefaultTimeZone fail to got IANA time zone
name , it do extra step to got  time zone name( 1 read current English time zone
name from registry ; 2 read current region name from registry(this also affect
conversion); 3 use ucal_getTimeZoneIDForWindowsID implement the conversion.)

please feel free to share your thoughts or feedbacks of this short version
patch.

best regards,

Yu Gong


________________________________
From: R-devel <r-devel-bounces at r-project.org> on behalf of yu gong
<yugong at outlook.com>
Sent: Wednesday, January 11, 2023 22:25
To: r-devel at r-project.org <r-devel at r-project.org>
Subject: [Rd] proposal for use ICU for timezone convertion on windows and a
draft patch

Dear All,

 Last week I found an issue about R Sys.timezone function on Windows and send
email to this list. This week , I read quite a few  docs and codes about windows
time zone to IANA time zone conversion. and found in unicode.org's  ICU
library has a function  ucal_getTimeZoneIDForWindowsID which do the time zone
convert. Microsoft also use it for time zone convertion in dotnet 6 and
7.(https://github.com/dotnet/runtime/blob/main/src/native/libs/System.Globalization.Native/pal_timeZoneInfo.c)
.

IMHO, use this function to implement time zone convert on windows is better than
R's hand-write conversion implementation, use the ICU , it will update the
conversion when unicode org upgrade the ICU library and no more hand-write
conversion code needed. Also RTools already included the ICU library,and R on
windows build link ICU lib on default  , so no additional library or tools need
for use this function for time zone conversion.

so I wrote a quick and dirty patch use the ucal_getTimeZoneIDForWindowsID, I
think maybe better discuss here before I send patch to bugs.r-project.org.
I put the patch on
https://github.com/armgong/misc-r-patch/blob/main/registryTZ-ICU.diff.

this patch mainly do following three steps : 1 read current English time zone
name from registry ; 2 read current region name from registry(this also affect
conversion); 3 use ucal_getTimeZoneIDForWindowsID implement the conversion.

please feel free to share your thoughts or feedbacks of this patch.  if it
doable ,then I will modify it and send it to bugs.r-project.org

best regards,
Yu Gong




        [[alternative HTML version deleted]]

______________________________________________
R-devel at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel

	[[alternative HTML version deleted]]