Patch for parsing dates and time

> Il 23/07/2012 15:47, Maarten van Beek ha scritto:
>> Indeed I didn't realise negative dates where possible I thought it was
>> an error. I'll fix that. However the following code shows the bug:
>>
>>         stream := '2011-02-01abcd' readStream.
>>         date := Date readFrom: stream.
>>         result := stream upToEnd.
>>
>>         result printNl.
>>
>> The suffix to the date is abcd, so after parsing the date the stream
>> should still contain abcd. However result only holds 'bcd'. This is
>> not tested in the current test cases. Time, Duration and DateTime have
>> the same bug.
>
> Ok, thanks.
>
>> Not being able to use position: is a problem though. When parsing this
>> time for example: '09:00::1234' you should read 09:00 as time and
>> leave ::1234 in the stream, but you have to look ahead 2 characters to
>> notice this.
>
> I think it's safe to consider this example ill-formed, and not care
> about it.
>
> Paolo

> On Mon, Jul 23, 2012 at 10:24:56PM +0200, Maarten van Beek wrote:
>> The attached patch accepts negative dates and doesn't use position:
>> anymore. I've removed the test cases that weren't possible to succeed
>> because of this. This includes some test cases for DateTime, where
>> DateTime skips over all separators to see if there's any usable
>> content left. These separators are thrown away.
>
> Hi,
>
> in general it is good to test what can be parsed and what can not
> be parsed. E.g. if someone wants to parse 09:00::1234 as 9:00 and
> it fails it easy to point him to the testcase that excludes this
> behavior, Pharo1.4 will throw an exception when parsing the above
> string as a Time, I think it would be acceptable to do the same.
>
> Somehow I am not able to quote the attachment from within mutt but
> you have some unused variables (e.g. you added 'pos' but don't make
> use of it).
>
> For Paolo I have included the diff of expout to stdout when using
> the right stream for parsing and then comparing what is left inside
> the stream. Maarten's test results are already a lot better.
>
> holger
>
>
> _______________________________________________
> help-smalltalk mailing list
> [hidden email]
> https://lists.gnu.org/mailman/listinfo/help-smalltalk
>

> Here's the new patch, I removed the unused variables and added extra
> tests which consisting of the tests that are expected to fail.
>
> Maarten van Beek
>
> On Tue, Jul 24, 2012 at 11:16 AM, Holger Hans Peter Freyther
> <[hidden email]> wrote:
> > On Mon, Jul 23, 2012 at 10:24:56PM +0200, Maarten van Beek wrote:
> >> The attached patch accepts negative dates and doesn't use position:
> >> anymore. I've removed the test cases that weren't possible to succeed
> >> because of this. This includes some test cases for DateTime, where
> >> DateTime skips over all separators to see if there's any usable
> >> content left. These separators are thrown away.
> >
> > Hi,
> >
> > in general it is good to test what can be parsed and what can not
> > be parsed. E.g. if someone wants to parse 09:00::1234 as 9:00 and
> > it fails it easy to point him to the testcase that excludes this
> > behavior, Pharo1.4 will throw an exception when parsing the above
> > string as a Time, I think it would be acceptable to do the same.
> >
> > Somehow I am not able to quote the attachment from within mutt but
> > you have some unused variables (e.g. you added 'pos' but don't make
> > use of it).
> >
> > For Paolo I have included the diff of expout to stdout when using
> > the right stream for parsing and then comparing what is left inside
> > the stream. Maarten's test results are already a lot better.
> >
> > holger
> >
> >
> > _______________________________________________
> > help-smalltalk mailing list
> > [hidden email]
> > https://lists.gnu.org/mailman/listinfo/help-smalltalk
> >
>