我其实想做一个电感式接近开关，

In that case, either approach should work, all you need to measure is changes in inductance, not inductance itself.

For example, you can periodically measure the inductance (lehman) or frequency (lc tank), and use that as basis for your comparison to see if the inductance has changed.

With this approach, temperature stability doesn't matter - unless temperature can change drastically between measurements (aka the target is a container of hot water, etc.).

我用LM311做了个LC振荡测电感的，

most mcus have internal opamp / comparators that can be configured for this purpose. no need for an external part.