Closed
Conversation
Previously we were using the unix assembly files with the SystemV calling convention. This predictably leads to crashes when running these jets on Windows, something that happens immediately when someone sends you a directed messaging packet. Here we switch to the Windows x64 calling convention assembly files that libblake3 so graciously provides.
Previously we were exclusively using unix assembly files with the SystemV calling convention. This predictably leads to crashes when running these jets on Windows, something that happens immediately when someone sends you a directed messaging packet. Here we switch to the Windows x64 calling convention assembly files that libblake3 so graciously provides.
Adds a compression handler, configured to support gzip and brotli for responses of non-trivial size. The library takes care of applying compression only on content types where this makes sense. For now, we're deferring to h2o's defaults, which is essentially a whitelist containing most common text-based mime types. We lightly adjust the `h2o_add_header_by_str` call so that h2o tokenizes our headers appropriately, letting it detect the content type.
Resolves #
Adds a compression handler, configured to support gzip and brotli for responses of non-trivial size. The library takes care of applying compression only on content types where this makes sense. For now, we're deferring to h2o's defaults, which is essentially a whitelist containing most common text-based mime types. We lightly adjust the `h2o_add_header_by_str` call so that h2o tokenizes our headers appropriately, letting it detect the content type. With this change, we see bytes-over-the-wire reduced as expected. A cache refresh of the groups web client on a ship with very little content (so, primarily .js file downloads) goes down from ~10mb to ~3mb transferred. (~30%) Filling a channel with small posts by the same author gives a "recent posts" json scry result of ~23kb, but only uses ~3kb over the wire. (~13%!) Happy to bikeshed the compression configuration here. I just picked values that seemed fairly middle-of-the-road.
The primary reason to upgrade lmdb here is the fact that latest master LMDB uses the NtCreateSection api to implement sparse files on windows. With the current windows binary every epoch immediately writes a 63 GB data.mdb file to disk which is less than ideal. Unfortunately there's no tagged release out with this fix, but this repository changes very infrequently and the LMDB guys know what they are doing so I determine the risk of upgrading here to be quite low. We also get LMDB/lmdb@82c5609 as a bonus with a fix for macos fsync so we can get rid of our patched mdb.c.
The primary reason to upgrade lmdb here is the fact that latest master LMDB uses the NtCreateSection api to implement sparse files on windows. With the current windows binary every epoch immediately writes a 63 GB data.mdb file to disk which is less than ideal. Unfortunately there's no tagged release out with this fix, but this repository changes very infrequently and the LMDB guys know what they are doing so I determine the risk of upgrading here to be quite low. We also get LMDB/lmdb@82c5609 as a bonus with a fix for macos fsync so we can get rid of our patched mdb.c.
Adds a `--gc-abort` flag which causes the runtime to crash if a leak is encountered, and replaces `--gc` in the testing bash script with that flag. Since the tests already run in CI with `--gc` and `-Dmem-dbg`, replacing that flag with `--gc-abort` would allow to catch memory leaks earlier in development
This `c3_free` call is left from before the mars-urth split when the struct was heap-allocated
Contributor
|
Thanks @dozreg-toplud, I cherry-picked the pertinent commits so that we don't duplicate commits from two merges of develop. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This should build on windows