Change Details

The `MergeHistory` module currently saves these log parameters: ``` '4::dest' => $this->dest->getPrefixedText(), '5::mergepoint' => $this->timestampLimit->getTimestamp( TS_MW ) ``` But it is not possible to obtain the exact ID of the target page, because the page might be renamed any time after the merge. Consider a client which synchronizes the content of a wiki into a local database: it might merge the revisions into a wrong page if it performed a simple lookup by the namespace and title. The logging table has a [log_page](https://www.mediawiki.org/wiki/Manual:Logging_table#log_page) field which is set to the target page ID on page moves. I propose to do the same with history merges.

The `MergeHistory` module currently saves these log parameters: ``` '4::dest' => $this->dest->getPrefixedText(), '5::mergepoint' => $this->timestampLimit->getTimestamp( TS_MW ) ``` But it is not possible to obtain the exact ID of the target page, because the page might be renamed any time after the merge. Consider a client which synchronizes the content of a wiki into a local database: it might merge the revisions into a wrong page if it performed a simple lookup by the namespace and title. The logging table has a [log_page](https://www.mediawiki.org/wiki/Manual:Logging_table#log_page) field which is set to the target page ID on page moves. ~~I propose to do the same with history merges.~~ Edit: Actually, the synchronization needs both IDs of the source and target page, so they should be both available in the log event at all times. The API currently provides `pageid` and `logpage` fields, but `pageid` is 0 when the source page is deleted after the merge. It would be probably easier to add the target page ID to the `log_params` field next to the target title.

The `MergeHistory` module currently saves these log parameters: ``` '4::dest' => $this->dest->getPrefixedText(), '5::mergepoint' => $this->timestampLimit->getTimestamp( TS_MW ) ``` But it is not possible to obtain the exact ID of the target page, because the page might be renamed any time after the merge. Consider a client which synchronizes the content of a wiki into a local database: it might merge the revisions into a wrong page if it performed a simple lookup by the namespace and title. The logging table has a [log_page](https://www.mediawiki.org/wiki/Manual:Logging_table#log_page) field which is set to the target page ID on page moves. ~~I propose to do the same with history merges.~~ Edit: Actually, the synchronization needs both IDs of the source and target page, so they should be both available in the log event at all times. The API currently provides `pageid` and `logpage` fields, but `pageid` is 0 when the source page is deleted after the merge. It would be probably easier to add the target page ID to the `log_params` field next to the target title.