On Mon, 17 Oct 2022, fractal.sum...@btinternet.com wrote:
Thanks very much for your response, and I take your point about the FTP
protocol in this respect. However, isn't the reason for having the optional
setting CURLOPT_IGNORE_CONTENT_LENGTH to give the caller a way to absolve
libcurl of providing this check and saying to it please just give me
everything the server gives you on the data socket. The onus then being on
the caller to check that they got what they wanted?
As the man page for CURLOPT_IGNORE_CONTENT_LENGTH says, this then allows
growing files to be supported, which is a nice feature, albeit one that
isn't as robust as for standard fixed transfers. The code to support this
option appears in the code base, but it looks to just have one small issue
that prevents it working. I took a clone of the latest version and put the
small fix in, which then allowed growing files to work for me locally.
Seems like a worthfile fix indeed.
Would you be able to make a pull-request out of your proposal? I would also
like to see a test added for this...
--
/ daniel.haxx.se
| Commercial curl support up to 24x7 is available!
| Private help, bug fixes, support, ports, new features
| https://curl.se/support.html
--
Unsubscribe: https://lists.haxx.se/listinfo/curl-library
Etiquette: https://curl.se/mail/etiquette.html
