Doctrine 2 - ORM
  1. Doctrine 2 - ORM
  2. DDC-752

Postpone Metadata Validation Until After Driver and loadClassMetadata Has Executed

    Details

    • Type: Improvement Improvement
    • Status: Resolved
    • Priority: Major Major
    • Resolution: Fixed
    • Affects Version/s: 2.0-BETA3
    • Fix Version/s: 2.0-BETA4
    • Component/s: Mapping Drivers
    • Security Level: All
    • Labels:
      None

      Description

      Right now the metadata drivers throw exceptions if certain metadata is not specified. While this is convenient for letting the user know about errors, it limits the extensibility of the ORM, specifically what you can do in the loadClassMetadata event. It also means that there is duplicated code in each of the driver implementations for validation.

      All validation of the metadata should be postponed until after the metadata driver has executed and also after the loadClassMetadata event has been fired. This way all metadata has been loaded before attempting to validate and the user is still warned about errors. This validation should still be ignored when loading from cache.

      Example use case:

      Currently, setting a @DiscriminatorMap is required when using inheritance mapping. The loadClassMetadata event provides a way to dynamically set discriminator maps from a config or a database without having to extend a metadata driver.

      The specific regression in functionality that this references is MappingException::missingDiscriminatorMap being thrown in the metadata drivers.

        Activity

        Hide
        Michael Ridgway added a comment -

        Added specific exception that has been causing issues for my use case. (MappingException::missingDiscriminatorMap)

        Show
        Michael Ridgway added a comment - Added specific exception that has been causing issues for my use case. (MappingException::missingDiscriminatorMap)
        Hide
        Roman S. Borschel added a comment -

        Sounds like a valid issue to me (and a regression).

        The metadata drivers should not do this kind of validation. If this validation needs to be done it should happen later either somewhere in ClassMetadata(Info) or the ClassMetadataFactory.

        Assigning this issue to Benjamin since I think he added this discriminator map validation in the drivers. I remember we talked about this but I did not properly recognize this issue either.

        Show
        Roman S. Borschel added a comment - Sounds like a valid issue to me (and a regression). The metadata drivers should not do this kind of validation. If this validation needs to be done it should happen later either somewhere in ClassMetadata(Info) or the ClassMetadataFactory. Assigning this issue to Benjamin since I think he added this discriminator map validation in the drivers. I remember we talked about this but I did not properly recognize this issue either.
        Hide
        Michael Ridgway added a comment -

        Committed a fix for my use case at http://github.com/mridgway/doctrine2/commit/9acfa8b7c3ea17f4ee75608deb4cb474a799353c

        This moves the discriminatorMap validation to ClassMetadataFactory::_loadMetadata

        Show
        Michael Ridgway added a comment - Committed a fix for my use case at http://github.com/mridgway/doctrine2/commit/9acfa8b7c3ea17f4ee75608deb4cb474a799353c This moves the discriminatorMap validation to ClassMetadataFactory::_loadMetadata
        Hide
        Benjamin Eberlei added a comment -

        fixed

        Show
        Benjamin Eberlei added a comment - fixed
        Hide
        Roman S. Borschel added a comment -

        I'm not sure but is it right that this validation in the CMF takes place before the loadClassMetadata event is fired? After all, this event is supposed to be used for dynamic completion of the metadata.

        Show
        Roman S. Borschel added a comment - I'm not sure but is it right that this validation in the CMF takes place before the loadClassMetadata event is fired? After all, this event is supposed to be used for dynamic completion of the metadata.
        Hide
        Benjamin Eberlei added a comment -

        i shouldn't code at late in the night, will fix it

        Show
        Benjamin Eberlei added a comment - i shouldn't code at late in the night, will fix it
        Hide
        Benjamin Eberlei added a comment -

        @Roman, what about the ID stuff? Can i just move the event before every CMF validation?

        Show
        Benjamin Eberlei added a comment - @Roman, what about the ID stuff? Can i just move the event before every CMF validation?
        Hide
        Roman S. Borschel added a comment -

        Maybe, but I would just try to leave the old order intact to not risk any BC issues. Simply moving the inheritance validation below the event should be enough for now.

        Show
        Roman S. Borschel added a comment - Maybe, but I would just try to leave the old order intact to not risk any BC issues. Simply moving the inheritance validation below the event should be enough for now.

          People

          • Assignee:
            Benjamin Eberlei
            Reporter:
            Michael Ridgway
          • Votes:
            0 Vote for this issue
            Watchers:
            1 Start watching this issue

            Dates

            • Created:
              Updated:
              Resolved: