Daily report (2026-10-04)

1. Summary

Work today has focused on two things. The NGINX server work is not finished. But progress has been made on that front. A computer upgrade has taken place. The latter took up quite a bit of time.

Yesterday’s NGINX work continued. It had been left at compilation. There had been issues with OpenSSL. This got eventually sorted out. But new issues awaited. FIPS providers are one huge rabbit hole.

Research into FIPS providers started then. This went through a number of package installations. Some of them were faily unrelated. Some packages were needed to OCR documents. These contained old information on FIPS providers.

This already took up a few hours. Time is at a premium. So a decision was made. A full system clean up was in order. Then a system upgrade would follow. And then package management utilities would be installed.

That took up a few more hours. Then came the NGINX work. It was fairly easy to install with a package manager. But that was it. The configuration is still tricky. More experience should make it easier to grok.

Current difficulties revolve around resolution. A certain URI pattern is easy to match. NGINX is also capable of forwarding. The issue is not hostname resolution. The issue is full request resolution.

The goal is to proxy requests against specific URIs. That should be possible through HTTP tunneling. The problem then becomes the target URIs. The client still requests a tunnel. But there’s need to lie to the client.

The client should then trust the connection. But that is TLS-encrypted. That is why the client can trust the tunnel. This will not become a project to break TLS.

2. Blockers

None for the main work. The proxy work is blocked on the above issue.

3. Plan for the week

The week will be spent on the Idris community tutorial. It will also be spent on the ctest extension. The latter will take up the first half of the week. Then there are other Rust matters to address in rust-lang/libc.

The proxy work will need some alternative solution. NGINX will be researched further. But it is potentially not suited for this task. One other possible solution is to alter kernel routing tables. That will wait until next weekend.