Raphael Geronimi
Raphael Geronimi
Hi, any news on this one? This prevents some feeds from being parsed
Yes there is one wrong behavior: the OSError object is not properly initialized with the errno information. So any user of the library catching OSErrors will obtain wrongly initialized objects....
Hi @alexmac , I could reproduce successfully your test case (i.e., with the results you did expect). You are correct, this is a subtle case about how constructors arguments are...
Indeed this _next_expiration field was used to perform fast removal of expired cookies, but a posterior commits deleted the line of code that was using it: https://github.com/aio-libs/aiohttp/blob/d65f5cb4751651c7c43ac3744c92cbd4b23eec04/aiohttp/cookiejar.py#L87 Since that line...
Same here - I use aiohttp for a large-scale crawler. It works wonder, excepted the time it spends processing the cookie jar. The code seems complex and, potentially, not optimized...
How many cookies were in the jar? I have several thousands (across a couple hundred domains)
Yep that's what I see. The cookie jar is not scalable, specifically due to the expiration logic. I didn't see this through `py-spy` but when switching the crawler from a...
In fact the entire logic could be switched in favor of a time-ordered expiry queue. That way, there is never any need to loop over every cookie. Only the ones...
Sounds great, I will wait for #7777 to be merged and then, depending on if it resolved this issue or not and the new capabilities it offers, make a proposal.
I understand, I will find the best path.