Alex Vondrak

The issue I see here is that the `'[ ... ]` syntax expands into a sequence which is `append!`ed to the parse tree, rather than a single object that's `suffix!`ed....

@erg Was thinking about it today, and wondered idly if there was a good way (or any way) the parser overhaul could help with this issue. Thoughts? Keep in mind...

> The parser will know the exact text that corresponds to every token, either in the image or on demand from the file-system, enabling it to write the code back...

I'm afraid I don't have any reliable minimal/local repro, as I mainly experience this in my full-blown production environment. I've gone ahead and made a minimal copy of my problematic...

The plot thickens (or maybe thinnens?). 🧐 While the `BLPOP` code from above hammered the reconnects (and thus the leaked connections), I just had another experience that leads me to...

> I'm not so surprised to be honest :/ Same :joy:

> The `synchrony` driver has been removed in 5.0.  ^ I'm allowed to post that, now that I'm no longer using synchrony in prod....

You're right, they aren't "real" user agent strings. The only two I've seen in the wild are the ones noted in my first comment. I used those UAs to extrapolate...

At any rate, I have trudged out some interesting examples from my logs - not exhaustive, but looking somewhat thorough so far (and still looking compliant with the formats tested...

This would be a nice to have, but I'm not sure it's really in the scope of this plug. At least the name would be misleading. 😛