412997 - Microseconds not normalized when they need to be

Status: RESOLVED FIXED

(Core Graveyard :: XForms, defect)

Description

[aaronr]

based off of a comment in bug 412028.

ValidateBuiltinTypeDateTime, ValidateBuiltinTypeDate, and
ValidateBuiltinTypeTime also need to take into account time zones.  They are
setting the second parameter to PR_ParseTimeString to PR_TRUE, meaning that the
info that is being passed in is in UTC, when in fact it isn't.  Perhaps
PR_NormalizeTime should be called first, before PR_ParseTimeString?

Comment 1

[aaronr]

testcase in the URL field.  This bug is why the last value is returned as 2 instead of 3.

Comment 2

[Merle Sterling]

Attachment: patch

ValidateBuiltinTypeDate/DateTime/Time must check if the time represents a GMT time so they can pass the correct value (true or false) for the 'default_to_gmt' parameter of PR_ParseTimeString.

Comment 3

[Doron Rosenberg (IBM)]

Comment on attachment 310333 [details] [diff] [review]
patch

>+  /**
>+   * Determine if an xsd:date, xsd:dateTime, or xsd:time string
>+   * represents a GMT (aka UTC) time.
>+   */
>+  PRBool IsGMT(const nsAString & aDateTime);
>+
> static void
> ReleaseObject(void    *aObject,
>               nsIAtom *aPropertyName,
>               void    *aPropertyValue,
>               void    *aData)
> {
>   static_cast<nsISupports *>(aPropertyValue)->Release();
> }

I think IsGMT belongs in nsSchemaValidatorUtils so we can reuse it later in the Utils methods if we need to.  Other than that, looks good.

Comment 4

[Merle Sterling]

Attachment: patch2

Moving isGMT function to nsSchemaValidatorUtils.

Comment 5

[aaronr]

Comment on attachment 310333 [details] [diff] [review]
patch

nit: add a comment to IsGMT so that we know that it assumes that aDateTime has been validated prior to being passed in.

with that, r=me

Comment 6

[aaronr]

checked into trunk for msterlin

Comment 7

[aaronr]

Attachment: patch for 1.8

patch for 1.8 branch

Comment 8

[aaronr]

checked into 1.8 branch for msterlin