On 01/12/2018 05:57 AM, Karel Zak wrote:
> On Thu, Jan 11, 2018 at 08:35:51AM -0500, J William Piggott wrote:
>>
>>
>> On 01/11/2018 04:01 AM, Karel Zak wrote:
>>> On Wed, Jan 10, 2018 at 09:00:23PM -0500, J William Piggott wrote:
>>>>
>>>>
>>>> On 01/08/2018 05:21 AM, Karel Zak wrote:
>>>>> On Tue, Jan 02, 2018 at 09:53:18AM -0500, J William Piggott wrote:
>>>>>
>>>>> I have no clue how many users care and read our ReleaseNotes, but
>>>>> important is that they have opportunity to do that and they have
>>>>> always time to adopt to changes. This is how I promised that this
>>>>> project will be maintained.
>>>>
>>>> Your response here puzzled me, because I could not remember you ever
>>>> making a pre-announcement of this nature before.
>>>
>>> 2.14 The losetup(8) '-s' option (introduced by util-linux-ng-2.13) is deprecated
>>> 2.21 The udev compatible output (-o udev) from blkid(8) is deprecated.
>>> 2.25 The "swapon --summary" output format is deprecated ...
>>>
>>> ... and many warnings about commands (mkfs, tailf, mount, last, ...) and
>>> features (cryptoloop).
>>>
>>> The command line options are our API, we don't do such changes often.
>>>
>>> I remember only one exception -- sfdisk, after rewrite some obscure
>>> DOS-era options have been removed.
>>
>> We're not talking about deprecating anything, which is why I said "of
>> this nature". We're not changing any API or removing any options.
>
> Well, my mistake. I have thought about "rename --julian to --ordinal"
> and errx(EXIT_FAILURE, ("use --ordinal, --julian has been deprecated."));
>
> This change is too aggressive without ReleaseNotes warning.
>
I agree, it is aggressive.
Having considered your comments, I realize that my submissions are
likely the wrong solutions. The reality is, the practice of calling
'ordinal Gregorian calendars' 'Julian calendars' is deeply entrenched;
it's even in the POSIX standard.
So, if we leave the --julian option as it is, what is the solution for
the name collision. Use some made-up name for a Julian calendar system
like --caesar. Sometimes Roman is used as an umbrella term for Roman and
Julian calendars combined. I don't think that's a good choice though,
because someone may get ambitious and implement output earlier than year
1; then it would be necessary to distinguish between Roman and Julian
calendars. Besides, to muddy the waters further by starting yet another
practice of using an incorrect name involving Julian just seems wrong.
I think the new option needs to be called Julian; so the only solution
left is to use a different name space. I propose the following:
* leave the current --julian option as is
* add one new option for setting the reform year:
--reform-year=<gregorian|iso|julian|1752>
I think this approach would also be better if the other 20 Gregorian
reform adoption dates are ever added. 24 arguments would be cleaner then
having an equal number of separate options.
Then to address whether to set the proleptic Gregorian calendar system
as the default: I think just stay with your original plan, only forget
the last bullet point with regard to implementing the code; I tried
including preparations for the switch in the files, but it's too messy. So:
* leave the current default using 1752 reform
* create the proleptic Gregorian option
* let Karel decide if/when/how/politics of changing the default
Maybe comments from closing the RedHat bug will help with that choice.
I mentioned before that I have more bug fixes queued, but didn't want to
make this patch set to large. Well, some of them influence handling of
the multiple calendar systems so I may add them to the next version of
this patch set.
Thoughts?
--
To unsubscribe from this list: send the line "unsubscribe util-linux" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
