MSP-Greg

> If deprecated protocols are not removed We have options to disable protocols, we haven't discussed whether to remove 1.1.1 specific code in MiniSSL when it's EOL. I've intended to...

I haven't been in the various specs for several weeks. See https://www.rfc-editor.org/rfc/rfc9110.html#name-content-length. The spec also refers to [Transfer-Encoding](https://www.rfc-editor.org/rfc/rfc9112#field.transfer-encoding), etc. The [Rack spec](https://msp-greg.github.io/rack/file.SPEC.html#label-The+Headers) specifies that header field names/keys should be lower...

While working on the test framework, some of the work involved removing the use of net/http. Its header API converts all header keys to lowercase. So, the API has one...

If think we should change them all to lower case. But, doing so may break any CI looking at header fields. So, it should either be a minor or a...

Also, see https://httpwg.org/specs/rfc9113.html#HttpHeaders

I just checked this issue, and it seems to behave the same for TLSv1.3 & TLSv1.2. See below, a negative ctx.timeout doesn't work with OpenSSL 3.2.0. JFYI, I patched a...

Looked at this a bit more, and OpenSSL 1.1.x allows storing a negative session timeout, as noted in the data below: * OpenSSL 1.1.1f Ubuntu 20.04 ``` ctx.timeout -2 session.timeout...

> Thank you for digging into this! That's about as far I feel comfortable going. I thought maybe throwing an error if the value is set to a negative (and...

@rhenium JFYI, MSFT/vcpkg just updated to OpenSSL 3.2.0. So, all Windows head builds are using 3.2.0, as MSYS2 previously updated to it. Happy Holidays / Merry Christmas Greg

@rhenium I just updated build tools locally, and noticed that all the Ruby head builds are using OpenSSL `OpenSSL 3.3.0 9 Apr 2024`...