Jose normally we do not summit code in the pane of the bug tracker, we commit a slice (click + slice in the monticello browser) in the inbox folder. Stef On Jun 7, 2013, at 9:12 PM, Stéphane Ducasse <Stephane.Ducasse@inria.fr> wrote:
tx I will look at it.
Stef
On Jun 7, 2013, at 8:57 PM, José Comesaña <jose.comesana@gmail.com> wrote:
Done.
I hope I have made it correctly. If not, please tell me.
Regards
2013/6/7 Stéphane Ducasse <stephane.ducasse@inria.fr> Here it is https://pharo.fogbugz.com/default.asp?10869
Stef
On Jun 7, 2013, at 8:17 PM, Stéphane Ducasse <Stephane.Ducasse@inria.fr> wrote:
Hi again.
I have one first test for #readFrom:pattern:. What can I do now to send it, where, to whom?.
I will create a bug entry for you and you create a slice with the test and the fix and you publish it in the pharo inbox.
Stef
Regards
2013/6/7 Sven Van Caekenberghe <sven@stfx.eu>
On 07 Jun 2013, at 14:28, José Comesaña <jose.comesana@gmail.com> wrote:
Answers in red.
Regards
2013/6/7 Sven Van Caekenberghe <sven@stfx.eu>
On 07 Jun 2013, at 13:26, José Comesaña <jose.comesana@gmail.com> wrote:
Thank you Sven.
I had made the fixes myself, creating a new class for dates (I don't need the complexity of Date just for determining if a date is past, I don't need date arithmetic either). I just wanted to inform, mainly because the bug seems to be still present in Pharo 3.0.
Anyway, your solution does not work for the 'd/m/y' pattern, although it does for 'd/m/yy'.
Depends on what a single y means, right ?
According to documentation, one y means ...year, no matter how many digits supplied... (my own words, not litteral). It means it will either accept ../../13 or ../../2013
Yes, that is how I understood it as well. My change adds 2000 only in the case of 2 y's unless I am mistaken. But I must admin I didn't test it.
I could write the tests you suggest, if you think they are interesting.
Yes, please do: with a good set of tests, we can discuss using concrete examples.
Sure I will. Will keep you informed!
Super.
Regards
2013/6/7 Sven Van Caekenberghe <sven@stfx.eu> Hi José,
On 07 Jun 2013, at 12:02, José Comesaña <jose.comesana@gmail.com> wrote:
Hello everybody.
I would like to add a new improvement to this thread.
I have noticed that Date class has a bug, or at least something to improve in readFrom:pattern: method. The initial comment states that "A year given using only two decimals is considered to be >2000.".
But this seems to be incorrect: if you evaluate
Date readFrom: '07/06/13' readStream pattern: 'd/m/y'
you get:
7 June 0013 instead of 7 June 2013.
I am working with version #20605
The fix would be quite easy:
year := (inputStream next: 2) asInteger
should be
year := 2000 + (inputStream next: 2) asInteger
A more important problem is that there are no users of #readFrom:pattern: in the system, not even unit tests. If we want to keep the methods, someone should start by writing a couple of unit tests.
Sven
Thank you all for your great work.
Regards.
2013/6/6 Camillo Bruni <camillobruni@gmail.com> can you save an image with the inspector opened on both x and y, x asTime and y asTime and provide a link to it? like that I can inspect it in all detail...
On 2013-06-06, at 17:51, dmacq <dmacq@instantiations.com> wrote:
I tried it again this morning and still had the problem.
This is interesting.
| x y | x := DateAndTime year: 1991 day: 196 hour: 20 minute: 5 second: 7. y := DateAndTime year: 1991 month: 7 day: 15 hour: 20 minute: 5 second: 7. x = y <------------------ True
But | x y | x := DateAndTime year: 1991 day: 196 hour: 20 minute: 5 second: 7. y := DateAndTime year: 1991 month: 7 day: 15 hour: 20 minute: 5 second: 7. x asTime = y asTime <------------- False
-- View this message in context: http://forum.world.st/DateAndTime-asTime-tp4691870p4691988.html Sent from the Pharo Smalltalk Users mailing list archive at Nabble.com.