Date: Mon, 1 May 2000 08:21:05 +0100 From: Andrew Josey To: jsm28@cam.ac.uk Subject: Defect Report response OGspecs 212 re: Single UNIX Specification Dear Joseph Enclosed is the response to your recent defect report. This response was developed and approved by the members of the Open Group Base Working Group. Please can you confirm receipt of this electronic mail message within 10 working days. Sincerely, Andrew Josey Chair, The Open Group Base Working Group -------------------------------------------- Response: The submitter has identified a real problem. Unfortunately, the problem is deeper than what is reported. There is currently nothing in a locale definition file identifying the adoption date of the Gregorian calendar and some places in the world still haven't adopted it. We do not recommend changing the behavior of historic implementations of the cal utility for UNIX '98 or UNIX 200x branding. But in XCU6 (currently in development), we should document that cal writes a calendar to standard output using the Julian calendar for dates from January 1, 1 through September 2, 1752 and the Gregorian calendar for dates from September 14, 1752 through December 31, 9999 as though the Gregorian calendar had been adopted on September 14, 1752. We should also add a note to the Future Directions section stating that a future revision of the standard may support locale specific recognition of the date of adoption of the Gregorian calendar. > >Resent-Date: 15 Apr 2000 00:50:42 -0000 >Date: Sat, 15 Apr 2000 01:48:35 +0100 (BST) >To: OGspecs@opengroup.org >From: jsm28@cam.ac.uk >Subject: Defect Report re: Single UNIX Specification >Resent-To: ogspecs@opengroup.org >Resent-From: ogspecs@opengroup.org > > > >Single UNIX Specification Defect Report: >----------------------------------------- > > From: Joseph Myers > Date: 2000 Apr 15 > >Address details: > >* This defect report is being made >personally, not on behalf of any >organisation or third party. * > >mailing address: Trinity College, >Cambridge, CB2 1TQ, UK > >Other contact methods not available: >use email. > >------------------------------------------------------------------------ > >Defect Report concerning (volume of the Single UNIX Specification): > >Commands & Utilities Issue 5 (XCU5) [UNIX 98] > >------------------------------------------------------------------------ > >Qualifier (e.g. error, omission, clarification required): > >2 > >Error=1 , Omission=2, Clarification=3 > >------------------------------------------------------------------------ > >References in document (e.g. page, clause, figure, and/or table >numbers or URL if applicable): > >http://www.opengroup.org/onlinepubs/7908799/xcu/cal.html > >------------------------------------------------------------------------ > >Nature of defect (complete, concise explanation of the perceived >problem): > > >The specification of the "cal" utility is unclear and possibly in >contradiction with traditional behaviour with regard to years >before the adoption of the Gregorian calendar. > >The traditional output of "cal 9 1752", in the POSIX locale, is >similar to: > > September 1752 > S M Tu W Th F S > 1 2 14 15 16 >17 18 19 20 21 22 23 >24 25 26 27 28 29 30 > >This, however, is not the "Gregorian calendar" required by Unix98. >It is a calendar of which two days are from the Julian calendar, >at that time in use in Great Britain and colonies, and the remainder >are from the Gregorian calendar, which was changed to at that date. >As such, the specification is unclear about whether, in the POSIX >locale, this output is permitted. Furthermore, the Gregorian calendar >was not instituted prior to 1582, rendering the requirements still >more unclear before that date: should the calendar be that commonly >if inaccurately known as the proleptic Gregorian (Gregorian rules >extended into the past)? > >Furthermore, the reference to "Gregorian calendar" also renders it >unclear whether the specification of, say, LC_TIME=en_GB, would allow >a display such as the one above to be given, showing a calendar that >is not strictly Gregorian, or whether the permitted effects of LC_TIME >are restricted to such matters as formatting and day and month names, >and potentially (given the unspecified format) any other information >about the days determined by the locale, but not what dates are >considered to be included in a given month. > >------------------------------------------------------------------------ > >Solution proposed by the submitter (optional): > >Specify that the calendar displayed may be Julian before a >locale-specific date, and may include locale-specific transition >behaviour between Julian and Gregorian. > >For the POSIX locale, specify the traditional behaviour of Julian >from 1752 September 2 and earlier, with the days September 3-13 >omitted, OR permit both this and proleptic Gregorian, the choice >being implementation defined (i.e., documented by the implementation). > > >------------------------------------------------------------------------ Approved: May 1 2000 ----- Andrew Josey The Open Group Director, Server Platforms Apex Plaza,Forbury Road, Email: a.josey@opengroup.org Reading,Berks.RG1 1AX,England Tel: +44 118 9508311 ext 2250 Fax: +44 118 9500110 GSM: +44 7974207557 SMS:aj@opengroup.org ** The Inaugural Realtime and Embedded Systems Forum meeting ** Austin, Texas, June 28th 2000 http://www.opengroup.org/realtime