Re: Re: [TSVN] Very long time to crawl WC.

> How it's currently done:
> * notification arrives
> * crawler has path of notification in queue
> * crawler starts with the first path in the queue
> * crawler fetches status of path
> * crawler checks queue for the same path it just crawled
> - if same path is in the queue again, remove it from the queue

Oops .. if there are no further mechanisms (timestamps etc.),
you will lose updates:

* start crawling $path
* shortly before crawling finished: modify already crawled part in $path
* add $path to queue
* crawling of $path is done
* remove all entries for $path
-> modification will not result in a status update

I suppose this issue is already handled somewhere
but if not, it possibly should ... ;)

Stefan.
Received on Thu Dec 1 20:32:29 2005