Summary: 85 instances, 81 unique

Text	Count
// TODO: Only ignore if this is from some past transaction	1
// TODO: are we intentionally piggyback the time value for flow control	1
// TODO: setSetting involves looping over all settings so the below actually	1
* TODO: we may consider exposing the additional debug data here in the	1
// TODO: with ByteEvent refactor, we will have to delay changing this	1
// TODO: sort out the TransportCallback for all the EOM handling cases.	1
// TODO: Pause reads and invoke infocallback	1
// TODO: send an RST if readBuf_ is non empty?	1
// TODO: avoid re-entrancy	1
// TODO: templatize the use of string as the value type; can instead use	1
TODO: I'd like to either extend this or add a variant that	1
// TODO: std::vector might be faster than std::list in the use case?	1
// TODO: move this logic down to session/codec	1
// TODO: Format directly into the IOBuf, rather than copying after the fact.	1
// TODO: remove tnx_ when not needed	1
// TODO: this one doesn't create trouble, but it's certainly not logging the	1
// TODO: handle the case where the priority node hangs out longer than	1
// TODO: deal with control streams in h2q	3
// TODO: evaluate whether using a list or default initializing vector with	1
// TODO: 400 a 1.0 POST with no content-length	1
// TODO: fill up other properties.	1
// TODO: only do this when stream is server-uni	1
* TODO: Reconsider default implementation here. If the handler	1
# TODO: Break up complex function	1
// TODO: ideally we should cancel the managed timeout when all the streams are	1
// TODO stream limit handling	1
// TODO: use a library function from boost?	1
// TODO: Audit this error code	2
// TODO: look into using headroom from data to hold the frame header	1
# TODO: audit that both source paths have the same inode number	1
// TODO: we need to emit this flag with the headers	1
// TODO: set Quic error when fbcode/quic/QuicConstants.h is OSS	1
// TODO: given the simple nature of TLCounter and that it is explicitly	1
// TODO T110695366: clean these up once clients have sufficiently migrated to	1
// TODO: instead of using DestructorGuard hacks to keep txn around,	1
# TODO: process the subdirectories of data_dir_name	1
// TODO share quic::AppToken class for serialization	1
# TODO: Introduce some sort of defaults config that can select by	1
* TODO: when HTTPSession sends a SETTINGS frame indicating a	1
# TODO: In the future it would be nice to support pre-compiling the source	1
// TODO: set a timeout?	1
// TODO: a possible optimization would be to discard datagrams destined	1
# TODO: On Windows we require the Python libraries as well.	1
# TODO: Our caller still won't be able to distinguish between an empty	1
// TODO: the codec will set this for non-H1Q	1
// TODO: should we treat msg as chunked like H2?	1
// TODO: consider changing API methods that take in an reference with versions	1
# TODO: we could just go ahead and build it here, but I	1
# TODO: We aren't extremely strict about error checking for one-value	1
* TODO: Separate method to do graceful shutdown?	1
// TODO: Duped id should trigger H3_SETTINGS_ERROR	1
// TODO: In STRICT_COMPAT, parseUrl.valid() is not checked, so URL.valid()	1
// TODO: 0-RTT settings	1
// TODO: set table size?	1
* TODO: optionally sort by qvalue descending	1
// TODO: if/when we support client goaway, reject stream if	1
# TODO: `mkscratch` doesn't currently know how best to place things on	1
// TODO: "trailers" may be received at any time due to the SPDY	1
// TODO: invoke socket.start() here	1
// TODO: CHECK(started_);	1
// TODO: qpackCodec_.setMaxUncompressed(setting.value)	1
// TODO support non http/https schemes	1
# system packages are installed.  TODO: For users not defined in the	1
// TODO: all the counters below are marked for deprecation.	1
// TODO: scale maxToSend by ratio?	1
// TODO: Below line will change for HTTP/2 -- just call a const getter	1
// TODO: replace the 2 booleans below with an enum "request method"	1
// TODO: referenence to head here is incompatible with baseIndex	1
// TODO staatic generateSettings	1
// TODO: We should examine if it's possible to actually support 1xx resp.	1
// TODO: ensure only 1 parse at a time on stack.	1
// PRIO_FIXME: this is done because of the corking business...	1
// TODO: or require the node to do it?	1
// TODO: we could send FIN only?	1
# TODO: generate a main executable that will invoke Python3	1
# TODO: There may be some way to approximate an availability	1
# TODO: Ideally, this function would inspect the current cgroup for any	1
// TODO: could call fn gated by std::enable_if	2
* TODO: when HTTPSession sends a SETTINGS frame indicating a	1
// TODO: Trim whitespace first	1
// TODO: use StringPiece for queryParams_ and delete splitNameValue()	1