[NLP-er]

Originally Posted by NLP-er

I notice that for some data cache is not used, translation is made again and again cache is populated. So there can be data duplication in DB. It is rare and didn't find out why yet (1 clue - working on that).

So ok - my clue is - hazard. When 2 pages with same not translated text are generating in same time it is possible that booth will find in cache nothing, booth will translate and booth will put it to cache. It can be easily resolved just by changing INDEX to UNIQUE. It will resolve data duplication, but...

I don't want to do this because afraid of errors when 2nd translation will be inserted. Because I cannot reproduce it in browser, I don't know how it will behave. Possible behavior is that 2nd translation just get error during insertion and just go further and everything is fine for end user (spider). But other possibility is that during that error 2nd page generation will fail (i.e. white screen). Note that even in this situation it can happen only once per page, and only when we have this hazard situation.

So does someone knows what really will happen (white page, error on page, or just silent mysql error and normal generation)?