>> It is still scary to read in libcurl's documentation that when pausing http/2 libcurl might buffer up to 32MB of data for each transfer!
>> Is there a "SETOPT" to control that? > Another way to reduce the caching would be to not pause that long while you read other transfers over the same connection... When writing a FileSystem (FS) with libcurl (eg curlftpfs), the FS does not know how long will the "pause" be since it depends on the behaviour of other application programs using the FS. At some point the FS might even need to reconnect if it was too long and the server closed the socket. > No, but we could make one. The issue is of course that the 32MB is the window you tell the server you have open, and reducing that window might have a performance impact. Thanks. So far I'm fine with the hardcoded value because I have not seen any "filelockers" use http/2 for the "storage" servers, but preparing such an option could be useful for the future or for other use cases. The issue again with FS, is if user applications open 100+ simultaneous files (soft ulimit is usually 1024), the FS could end up eating Gigabytes of memory! Cheers Alain ------------------------------------------------------------------- Unsubscribe: https://cool.haxx.se/list/listinfo/curl-library Etiquette: https://curl.se/mail/etiquette.html
