Conversation
Bump XDR pin to stellar-xdr 03cbf40c and regen for millisecond-resolution close times. Enables MS_CLOSE_TIME on the next channel only (curr picks up the pin's already-live CAP-0083 STELLAR_VALUE_EMPTY_TX_SET arm, no CAP-0088 content). Also pins XDRGEN_COMMIT ahead of xdrgen#233, which dropped the JavaScript generator upstream, and hand-patches types/*.d.ts since dts-xdr's raw transform output no longer matches this repo's committed module wrapper.
|
|
|
|
Changes
03cbf40cand regen; enablesMS_CLOSE_TIMEon thenextchannel only (StellarValueSignedMsValue/StellarValueProposedMsValue+TimePointMilliseconds).curralso picks up the pin's already-live CAP-0083STELLAR_VALUE_EMPTY_TX_SETarm (no CAP-0088 content), since both channels share one.xsnapshot.stellar-xdr xfile preprocess,scripts/post-process-generated.pyconst-inlining, dropped-it,node:lts-alpine) — none of this existed onmasteryet.XDRGEN_COMMITtod54959f(just before Remove the JavaScript code generator xdrgen#233 removed the JS generator upstream) somake generatestill works.types/curr.d.ts/types/next.d.tsfor the new arms — dts-xdr's raw transform output no longer matches this repo's committed module wrapper (missing header/export/xdrHiddenhack), so a full regen isn't usable; patched surgically instead, mirroring the CAP-0085 precedent.Out-of-scope observations
xdrgenmasterentirely — breaksXDR_BASE-style regen for every consumer of this Makefile, not just this CAP. Worked around here via theXDRGEN_COMMITpin; flagging upstream separately.src/transform.jsoutput no longer matches this repo's committedtypes/*.d.tswrapper (header comment,import { Operation } from './index',export {},xdrHiddennamespace) — those aren't produced by the automated transform and must be hand-maintained. Not fixed here (out of scope); flagging for future regens.Downstream: none yet in this stack.