httpuv
httpuv ships almost nothing, which is what you want from the web server under Shiny
A side-by-side editorial comparison of promises and TypeDB — release velocity, themes, recent moves, and the top alternatives to consider.
promises drops R6 for plain environments and wires async work into OpenTelemetry
promises supplies the async primitives behind Shiny and plumber. After years of small domain and future() fixes, the last two releases changed both what a promise costs and what it can report: 1.4.0 corrected nested domain ordering and added the first OpenTelemetry integration, and 1.5.0 replaced the R6 promise class with a classed environment while stabilizing the tracing API.
TypeDB is making the query path cheaper and the schema finally editable.
The 3.11 and 3.12 lines have been a run of server-side correctness and operability work: a driver compatibility floor that rejects anything older than 3.11.0, connect hints printed at startup, pre-created UUIDs for users so a distributed deployment can share state, transaction close made synchronous instead of fire-and-forget, and RocksDB cache and write-buffer limits exposed as configuration. The newest release splits the old compilation cache into separate parse, translation and compile stages, and adds type renaming through redefine.
promises supplies the async primitives behind Shiny and plumber. After years of small domain and future() fixes, the last two releases changed both what a promise costs and what it can report: 1.4.0 corrected nested domain ordering and added the first OpenTelemetry integration, and 1.5.0 replaced the R6 promise class with a classed environment while stabilizing the tracing API.
Two things are being fixed at once. Promise domains - the mechanism that carries context across async boundaries - have been the source of nearly every bug in this window, and they are now being used deliberately as the vehicle for tracing rather than just for error handling. Meanwhile the object itself is being made cheap enough to create in volume.
Expect the OpenTelemetry surface to lose its experimental label and pick up the same span coverage Shiny would need end to end; the deprecated ospan function names should be removed in a following release.
The 3.11 and 3.12 lines have been a run of server-side correctness and operability work: a driver compatibility floor that rejects anything older than 3.11.0, connect hints printed at startup, pre-created UUIDs for users so a distributed deployment can share state, transaction close made synchronous instead of fire-and-forget, and RocksDB cache and write-buffer limits exposed as configuration. The newest release splits the old compilation cache into separate parse, translation and compile stages, and adds type renaming through redefine.
Two threads are visible. One is making the engine predictable for operators — memory budgets, transaction guarantees, explicit version floors. The other is reducing per-query cost now that the given stage makes string-identical queries common, which is why parsing is separated from translation and can happen without a transaction. Type renaming is the first real schema-evolution affordance in this window, and it arrived alongside the caching work rather than as its own release.
Expect the cache split to be followed by invalidation tuning, since translation and compile caches flush on schema commits and statistics drift, and further redefine-based schema evolution now that renaming works.
Other DevOps products tracked by Sparkpulse, ranked by recent ship velocity. Each card links to a full editorial trajectory and lets you pivot into a head-to-head comparison with either promises or TypeDB.
httpuv ships almost nothing, which is what you want from the web server under Shiny
withr rebuilt defer() on base R's own on.exit(), and got much faster for it
htmltools spends three releases walking back its own fill-layout opinions
xml2 relicensed to MIT, then moved its hot accessors from S3 into C
httr's last releases are housekeeping - the package now points users at httr2
magrittr rewrote its pipe in C, then base R shipped one and the releases stopped
See all promises alternatives → · See all TypeDB alternatives →
Latest ship moves from both products, interleaved chronologically. ⚡ = editorial spark.
They serve adjacent needs but don't currently overlap on shipped themes. TypeDB is currently shipping more aggressively (velocity 2.5 vs 0.0), with 0 editorial sparks in the last 30 days against 0. See the at-a-glance table above for a side-by-side breakdown of velocity, recent sparks, and editorial themes.
Sparkpulse doesn't pick a winner — we score release velocity, not feature parity. TypeDB is currently shipping more aggressively (velocity 2.5 vs 0.0), with 0 editorial sparks in the last 30 days against 0. For your specific use case, the alternatives sections above list other DevOps products to evaluate alongside.
Top promises alternatives in DevOps are ranked by recent ship velocity. Browse the "promises alternatives" section above for the current picks, or visit /alternatives/promises for the full list with editorial commentary on each.
Top TypeDB alternatives in DevOps are ranked by recent ship velocity. Browse the "TypeDB alternatives" section above for the current picks, or visit /alternatives/typedb for the full list with editorial commentary on each.