Re: Attn: clj-time users - possible breaking change being considered!

2012-02-06 Thread Cedric Greevey
On Mon, Feb 6, 2012 at 3:43 PM, Caspar Hasenclever wrote: > In other words, the value of (date-time 1970 1 1) should not, in my > opinion, depend > on whether it is run on my machine or yours. This is an important, valid point. Probably I should amend my earlier suggestion from a switch that use

Re: Attn: clj-time users - possible breaking change being considered!

2012-02-06 Thread Sean Corfield
Good feedback Casper, thanx! On Mon, Feb 6, 2012 at 12:43 PM, Caspar Hasenclever wrote: > As Avi points out on the github issue discussion, this change would > best be done throughout, i.e. wherever a DateTime instance is created, > otherwise one would end up with surprising behaviour (default of

Re: Attn: clj-time users - possible breaking change being considered!

2012-02-06 Thread Caspar Hasenclever
As Avi points out on the github issue discussion, this change would best be done throughout, i.e. wherever a DateTime instance is created, otherwise one would end up with surprising behaviour (default of UTC in date-time, default of JVM default time zone in formatter). That being said, I would arg

Re: Attn: clj-time users - possible breaking change being considered!

2012-02-06 Thread Sean Corfield
On Mon, Feb 6, 2012 at 11:17 AM, Cedric Greevey wrote: > Maybe there should be a *switch* for that ... A good suggestion as a possible compromise to allow both defaults. Right now, my default position is to not change anything unless enough folks indicate a desire for default time zone per Avi's

Re: Attn: clj-time users - possible breaking change being considered!

2012-02-06 Thread Cedric Greevey
On Mon, Feb 6, 2012 at 2:09 PM, Sean Corfield wrote: > Also worth noting is that Avi pointed out that (now), > (today-at-midnight), (epoch) etc all use UTC instead of the default > time zone so this is a broader philosophical point of whether clj-time > should continue to use UTC as its default or

Re: Attn: clj-time users - possible breaking change being considered!

2012-02-06 Thread Sean Corfield
On Sun, Feb 5, 2012 at 5:23 PM, Avi Flax wrote: > What about: (formatter "fmtstr" (default-time-zone)) ? Ah yes, but if that's the normal desired behavior that's an ugly default compared to (formatter "fmtstr")... Also worth noting is that Avi pointed out that (now), (today-at-midnight), (epoch)

Re: Attn: clj-time users - possible breaking change being considered!

2012-02-06 Thread Avi Flax
On Feb 5, 7:31 pm, Sean Corfield wrote: > and there's no > easy / obvious way to create a formatter with the default timezone, > without dropping down to the underlying Java. Not that I want to weaken my own case, but isn’t there? What about: (formatter "fmtstr" (default-time-zone)) ? That said

Attn: clj-time users - possible breaking change being considered!

2012-02-05 Thread Sean Corfield
Avi Flax recently pointed out that (formatter "fmtstr") creates a formatter in UTC whereas the underlying Joda Time library uses the default timezone when you do DateTimeFormat.forPattern(): https://github.com/seancorfield/clj-time/issues/8 Given that a two-argument version exists to specify a ti