This feels like neglected software to me. It has a strong odor of a product that the owning team does not consider a major priority and takes an ‘is it working well enough?’ approach to issues rather than an ‘is it working well?’ approach.

You can kind of identify other software systems/features of other tech products that are neglected in this same way: it’s the API that flakes out, or the UI that missed the update/reskin and probably runs on neglected legacy backends, or whatever.

And before the inevitable indictment of Google engineering’s failure to consider non-Google-scale systems: that might be the root of the neglect, but it could just as easily have a dozen more mundane reasons, from “team just doesn’t like working on it/hates the code” to “some silly operational constraint makes it very hard to make the needed changes to the system”.

Like, sure, all those things are the maintainers’ (Google’s) fault, and certainly fixable given the maintainers’ resources. This isn’t a defense of Google; just a statement that the specific reasons for goproxy sucking might not be in the list of stereotypical sweeping indictments people often presume about $GOOG.

> neglected software

I think it's sort of sad that this happens.

There is some really important software that - with a little talented person attension - could make everyone's life better.

I think of "make" and how useful it is, and how tiresome it is for everyone. I think taking effort out of cryptic makefiles and putting a little effort into make itself could really make lives easier. (one easy example: tab handling in makefiles)