[TYPO3-core] RFC #8522: Bug: Datetime shifts on copy and localize (timezone related)

[Steffen Kamper]

"Ernesto Baschny [cron IT]" <ernst at cron-it.de> schrieb im Newsbeitrag 
news:mailman.1.1211880423.30213.typo3-team-core at lists.netfielders.de...
> This is an SVN patch request.
>
> Type: Bugfix
>
> Bugtracker references:
> http://bugs.typo3.org/view.php?id=8522
>
> Branches:
> trunk and TYPO3-4_2
>
> Problem:
> Input-Fields with eval "date" or "datetime" will get its value shift by
> "server-timezone" minus "UTC" when copying the record or localizing the
> record.
>
> How to reproduce:
> Install the "user_datetest" extension attached to the bug report. This
> will just create a new table with all possible datetime fields avaliable
> to test and also make them localizable. Follow the steps:
>
> 0) make sure your server is not running on UTC, else the bug will not
> show up. Any "german server" should be ok (UTC+2 currently).
> 1) create a "Date Tests" record, tick every checkbox to get current date
> and times in all fields.
> 2) save the record.
> 3) copy the record (e.g. using the clipboard): Result without the patch
> are shifted datetimes.
> 4) localize the record to another language: Result without the patch are
> shifted datetimes in the localized record.
> 5) now apply the patch.
> 6) repeat steps 3) and 4), watch the results.
> 7) come back to the list give a +1.
>
> Solution:
> Attached patch. Thanks to lee perry for the original bug report.
>
> Notes:
> The problem comes from the fact that TCEmain "checkValue"
> transformations were being applied on already transformed data (from the
> database) even in case of a copy (which is also used on "localize").
> Until now there was no problem, because all checkValue transformation
> would do no harm on already transformed input (e.g. trim on trimmed
> string does nothing, "upper" on already upper cased string is harmless,
> etc), but it turned out to be a problem on the transformation needed for
> datetime fields (shifting the INPUT coming from the client, which is
> expected as UTC to server-TZ timestamp).
>
> So the solution is not to apply any checkValue in case of COPY or
> LOCALIZE (which uses COPY), which is what the patch does.
>
>
> Cheers,
> Ernesto
>
Hi Ernesto,



thx for taking care.

+1 by reading and testing (tested with localize)

vg Steffen