Summary: 122 instances, 109 unique

Text	Count
-- TODO: use a short string builder	1
// TODO: We can drop the current's fact ownership info here - could shrink	1
-- FIXME: This is a terrible way to get the terminal size but we don't	1
1: Repo stacked; // TODO remove?	1
// TODO: Should `completeOwnership` work with the trie rather than a	1
-- TODO: different error?	1
// TODO: replace canonicalize() with std::path::absolute() when	4
-- TODO: Make it better. Seems to be prone to errors	1
-- TODO: We probably want to read the repo config from a file, or put it in a	1
// TODO: We don't handle class-scope UsingDecls for now as we'd have to	1
// TODO: this is wrong	1
// TODO: It's not clear if/when this is used instead of	1
-- TODO: This isn't very efficient	1
-- TODO: Start substituting the next batch while the current batch	1
-- TODO: make this atomic if necessary	2
// TODO: What should we do if it's invalid?	1
-- FIXME: There is a tiny race here where we might fail in a weird way	1
-- TODO: we can generalize this to return more traces	1
// TODO: Do we want to resolve symlinks ourselves instead of using	1
// TODO: is this always public?	1
-- TODO: this could go away if we could get Thrift to use ByteString	1
/// TODO: We want to share one capacity between all db caches.	1
// TODO: initialize this lazily	1
return decl;    // TODO: is this right?	7
-- TODO: The queries shouldn't be hardcoded here and there should be a lot more	1
-- TODO: Decide anonymousNamespace | intentionallyEmpty | missingKey	1
-- TODO: Support logging multiple repos	1
-- TODO: we should probably have MatchRefs [Fid] for	1
-- TODO: avoid copy	1
-- FIXME: This is a terrible hack to ensure we don't remove everything when	1
-- TODO:	1
-- FIXME: This is a truly terrible hack.	1
-- TODO: this could be replaced with a single bulk fetch and Decl	1
-- TODO: there is a bug with const	1
TODO: we should also filter out Pids of predicates that were	1
// TODO: Validate UTF-8?	1
-- TODO: if this is a trusted fact ID (i.e. not supplied by	1
// TODO: defer this, see comments in evict	1
// TODO: This is a temporary hack for backwards compatibility, add proper	1
-- TODO: should take latest configuration repo list	1
// TODO: Store the set in the DB and assign it to the current fact.	1
, idsRef :: {-# UNPACK #-} !(IORef [Fid])  -- TODO: use a Storable Vector	1
-- TODO: exception handling behavior	1
// TODO: verify that predicates have the expected types	1
-- TODO: extend this to include all selectors with ident	1
// TODO: remove?	1
// FIXME: probably do this before f	1
-- TODO: snapshot the env, rather than passing in the mutable fields	1
-- TODO: Support logging multiple dbs	1
-- TODO: Decide anonymousNamespace | intentionallyEmpty | missingKey	1
// TODO: make this into an iterator over facts	1
-- TODO: can be optimised if necessary	1
-- TODO: What if someone is already deduplicating another batch? Should we	1
TODO:	1
// TODO: If we're doing a prefix seek and demand requires values, this	1
// TODO: can this ever happen? or will it always be a MemberExpr?	1
/// TODO: We probably want to either always return a path that's relative to	1
-- TODO: What do we do in this case?	1
-- TODO: reimplement based on dStringRef	1
-- TODO: More knobs.	1
Cxx.FunctionName_key_literalOperator _ -> True -- TODO: check it	1
-- TODO: later we will store the slice in the stacked DB, and read it	1
-- TODO: avoid the encodeUtf8 (name should really be a	1
// Used to identify versions of a predicate (TODO: merge with schema.thrift)	1
-- FIXME: This can loop if the generated code contains infinite loops	1
-- TODO: Clean up work queue - this will happen as workers	1
, Logger.setUploadDestination "<TBD>"     -- TODO: not sure how to	1
Just (x, classifyPattern lhsScope key) -- TODO lookupScope here is wrong	1
/// TODO: Say we cache the result of idByKey but then only ever do typeById	1
// TODO: log this to Scuba if it turns out to happen a lot	1
-- TODO: make this atomic if necessary (cf. renameat2)	1
// TODO: try to vectorise and/or use lookup table	1
// TODO: should we ignore deleted functions or have some info about them?	1
-- TODO: implement buffered downloads in Manifold client	1
-- TODO: matching like this is a bit janky.	1
// TODO: Try adding a fixed-size LRU (or LFU?) cache for set unions?	1
// TODO: complete	1
] -- TODO: List missing dependencies	1
-- ownership. (TODO: support ownership in the memory backend and	1
// TODO: Remove this copy	1
-- TODO: scoped searches for Objc methods	1
// FIXME: what *should* we do if the end of the range is in a different	1
-- TODO : do the plumbing via a class rather than function composition	1
-- TODO: This will loop if we have recursive typedefs but we don't allow	1
-- TODO: Move to Configerator (T64608954)	2
-- TODO: reimplement based on dByteStringRef	1
-- TODO: this is lame - it's not clear what to do if we can't	1
// FIXME: This isn't quite right, the qualified name should be	2
, Fid(..)   -- TODO: Fid isn't very useful, might want to tidy up the API here	1
// FIXME: This is quite expensive and not always used. We should do this	1
-- TODO: move into facebook/ ?	1
// TODO: set the right context for constructor initialisers and exception	1
-- TODO: this should pass the configured repo list through	1
-- TODO: try the other layers when this is a stack	1
// TODO: This is very hacky, make it better.	1
// FIXME: complete	1
-- TODO: read byte rather than Nat if arity <= 127	1
/// TODO: We probably want to remove facts which haven't been used for some time	1
-- TODO: Better statistical analysis.	1
-- TODO: ownership	2
Cxx.FunctionName_key_conversionOperator _ -> True -- TODO: check it	1
-- FIXME: What we're currently doing still isn't enough. Consider this:	1
// TODO: We implement substitution via the typechecker for now but it we	1
// TODO: implement	1
/// TODO: Switch to a variant of the Clock algorithm or something similar which	1
// TODO: We should probably cache this rather than recomputing every	1
// TODO: We aren't cross-referencing local variables for now but should	1
// TODO: We probably want to look up in stacked first but the current	1
-- TODO: this shouldn't really go under schema/thrift	1