> I would think they are, but I can't really comment on generalisations. I can't either, that's based on experiment in various scenarios, so I used "seems". But this is the most reasonable and understandable approach for now based on the outputs.
> Please elaborate. How is it expensive? The most intuitive one I would say is the overhead of print. If we need to differentiate the error's source by printing, or even dumping the message, that works well on single transaction, but if we've got hundred's of thousand's of transaction per hour, printing or dumping could significantly cause bad performance, that's why I called it "expensive". > If you want the data you need to have it enabled. Yes, that's also the conclusion based on what I've found. So I gave up trying to get all the debug message and changed to do the differentiation in another way, and now it's solved. Thank you for reply. Best Kuan-Sheng
------------------------------------------------------------------- Unsubscribe: https://cool.haxx.se/list/listinfo/curl-library Etiquette: https://curl.haxx.se/mail/etiquette.html
