March 2017 Archives by author

• Messages sorted by: [ thread ] [ subject ] [ date ]
• More info on this list...

Starting: Wed Mar 1 00:04:05 CET 2017
Ending: Fri Mar 31 17:14:06 CEST 2017
Messages: 250

• [master] bffb15e First stab at hit-for-pass documentation   Dridi Boukelmoune
• [master] 9837045 Send a PROXY v1 line with backend probes   Dridi Boukelmoune
• [master] 1087f15 Send PROXY v2 LOCAL header with probes   Dridi Boukelmoune
• [master] ba721c2 Tautological assert   Dridi Boukelmoune
• [master] 2b94cd5 v1 actually forwards traffic for v2   Dridi Boukelmoune
• [master] 3a27f6f Missing logexpect name   Dridi Boukelmoune
• [master] 17a456b Harmonize name checks in varnishtest   Dridi Boukelmoune
• [master] b101ec4 Needless branch   Dridi Boukelmoune
• [master] 4c9b097 Move setenv to vtc_fatal   Dridi Boukelmoune
• [master] 2d927b1 Don't protect the barriers list with a lock   Dridi Boukelmoune
• [master] dd17e5c Constify   Dridi Boukelmoune
• [master] b4bd652 The file descriptor is not leaked   Dridi Boukelmoune
• [master] 06f9176 A hash_always_miss can override a hit-for-pass   Dridi Boukelmoune
• [master] 213a251 A hash_always_miss can override a hit-for-pass   Dridi Boukelmoune
• [master] 3b70eaa Revert "Cargo-culting for glorious victory for autocrap hegemony"   Dridi Boukelmoune
• [master] 448d26c Fix VPATH builds   Dridi Boukelmoune
• [master] 0e4174c Fail name checks early in varnishtest   Dridi Boukelmoune
• [master] 5c8b790 Feed automake a pkgdatadir   Dridi Boukelmoune
• [master] ffa620d bah, trusty does not have nghttp2   Dridi Boukelmoune
• [master] ef4dc6e Lck_ operations are limited to varnishd   Dridi Boukelmoune
• [master] 7c328f9 Make miniobj work downstream   Dridi Boukelmoune
• [master] 35ac814 Update fetch FSM with return(pass)   Dridi Boukelmoune
• [master] 89705bd Warn about global state in VMODs   Dridi Boukelmoune
• [master] 752a002 Clarify   Dridi Boukelmoune
• [master] 70d37c5 Breaking change   Dridi Boukelmoune
• [master] 44f9423 Remove euphemism   Dridi Boukelmoune
• [master] 3857621 Mention BACKEND comparison now fully working   Dridi Boukelmoune
• [master] c31390c The `process` command was introduced in 5.0   Dridi Boukelmoune
• [master] 8be1d6b Suggest -f '' with -I in varnishd(1)   Dridi Boukelmoune
• [master] a96e465 Whitespace OCD   Dridi Boukelmoune
• [master] 7ef9b11 Mention that varnishd -x outputs RST   Dridi Boukelmoune
• [master] 4c9f9e0 Fix rendering of varnishd -? in sphinx   Dridi Boukelmoune
• [master] 100adab Help sed understand what the expression is   Dridi Boukelmoune
• [master] fc25bb5 Revert "Help sed understand what the expression is"   Dridi Boukelmoune
• [master] 85658ab Revert "Fix rendering of varnishd -? in sphinx"   Dridi Boukelmoune
• [master] 0ae899d Make a bigger deal of VXID in VSL queries   Dridi Boukelmoune
• [master] c207078 Describe VMODs as VCL modules   Dridi Boukelmoune
• [master] fca8307 More VRT changes   Dridi Boukelmoune
• [master] 0b84709 Avoid platform-dependent documentation   Dridi Boukelmoune
• [master] 0c1408c Revert "Describe VMODs as VCL modules"   Dridi Boukelmoune
• [master] b0190d0 Ignore failures from the osx job on Travis CI   Dridi Boukelmoune
• [master] 5cc2916 Add ${pkgsysconfdir} and ${vcldir} to varnishapi.pc   Dridi Boukelmoune
• [master] f62e60d Polish: s/varnishconfdir/pkgsysconfdir/   Dridi Boukelmoune
• [master] 94c9a54 Package-wide and per-VMOD vcldir support in varnish.m4   Dridi Boukelmoune
• [master] 39e65f1 Polish: maintain PATH in the doc example   Dridi Boukelmoune
• [master] 3fbeda7 Polish: do the --prefix inheritance in one place   Dridi Boukelmoune
• [master] a04a489 Changelog update for #2251   Dridi Boukelmoune
• [master] 52099d4 Don't append newline for varnishtest -p   Dridi Boukelmoune
• [master] bc2ea2c Support multi-lib systems for out-of-tree builds   Dridi Boukelmoune
• [master] 99b1a62 Check backend existence when reading proxy_header   Dridi Boukelmoune
• [master] 5087989 Teach http_CollectHdr to use arbitrary separators   Dridi Boukelmoune
• [master] d26a828 Collapse Cookie headers in HTTP/2   Dridi Boukelmoune
• [master] 58b2d30 Grow a separator argument in std.collect   Dridi Boukelmoune
• [master] eadcc45 Ignore leading spaces in http_CollectHdr   Dridi Boukelmoune
• [master] 99efebb Coverage for std.collect   Dridi Boukelmoune
• [master] 84e98d8 Improve documentation for std.collect   Dridi Boukelmoune
• [master] 91ee053 Merge pull request #2216 from rezan/fix/age_rounding   GitHub
• [master] 3767269 How to report signal panics   GitHub
• [master] 84b76e8 try to make coverty happy   Nils Goroll
• [master] 2591b9a fix the std.getenv example   Nils Goroll
• [master] 428996a fix and improve std.late_100_continue documentation   Nils Goroll
• [master] d466f8e vcc: support negative DURATION constants   Nils Goroll
• [master] a8b453c Allow EPIPE in VTCP_Check()   Nils Goroll
• [master] 528fd74 sync VRB_Ignore() comment with reality   Nils Goroll
• [master] 11fe8b0 I want to find this vtc when grepping for waitinglist   Nils Goroll
• [master] 3e015a1 Do all of the vfc setup and init in one place and only when needed   Nils Goroll
• [master] cf6a157 attempt to get more helpful bug reports   Nils Goroll
• [master] acd4c29 no debuginfo packages on packagecloud.io   Nils Goroll
• [master] b758894 try to stabilize this test for make -j80 check   Nils Goroll
• [master] acd2198 Fix race in the VSL segment counter initialization   Martin Blix Grydeland
• [master] 6cb0c66 Add parenthesis to clearify code   Martin Blix Grydeland
• [master] ad8e8f8 Do not update mgt's mirrored VCL temp state on child temp state failures   Martin Blix Grydeland
• [master] 744e55d Log VSM/VSL errors in varnishtest   Martin Blix Grydeland
• [master] 110d073 Add a vsl_catchup varnishtest function   Martin Blix Grydeland
• [master] 5dd70af Set the go-cold timer on newly loaded warm auto VCLs   Martin Blix Grydeland
• [master] 51ba68f Prepare for 5.1.0 final   PÃ¥l Hermunn Johansen
• [master] de38712 Prepare for 5.1.1 final   PÃ¥l Hermunn Johansen
• [4.1] 5030448 add std.getenv()   PÃ¥l Hermunn Johansen
• [4.1] 2bc20ee Repair test case m00026.vtc   PÃ¥l Hermunn Johansen
• [4.1] 3c1c970 Dramatically simplify VEV   PÃ¥l Hermunn Johansen
• [master] b2f32cd Try to get http->protover into some kind of consistent state so we do not send HTTP/2.0 headers to backends on pass from HTTP/2.0 clients   Poul-Henning Kamp
• [master] 213939a Don't issue Unset VSL records for the three top-line fields.   Poul-Henning Kamp
• [master] 62d4ce7 Also try to ditch thread pool with poll waiter   Poul-Henning Kamp
• [master] e5b99ec Split H2 session from protocol/frame processing   Poul-Henning Kamp
• [master] 9138957 Handle bogus ping -ack packets   Poul-Henning Kamp
• [master] bc38ebe More nitpicking while waiting in train.   Poul-Henning Kamp
• [master] 0ce6565 add needless returns   Poul-Henning Kamp
• [master] e5f8204 Make Continuation frames non-panicky.   Poul-Henning Kamp
• [master] 89af3f9 Some of the special cases of RST_STREAM   Poul-Henning Kamp
• [master] e9f32d6 HTTP2 is too complex, cite RFC's by line-numbers to give people a chance.   Poul-Henning Kamp
• [master] c0330b4 Tabelize a lot of rfc7540 "MUST" error handling   Poul-Henning Kamp
• [master] 17ad98c With table-driven MUST handling, some checks become simpler.   Poul-Henning Kamp
• [master] 3369999 The proto/session split ended up with three duplicated functions, GC them.   Poul-Henning Kamp
• [master] 6207b70 Fix a self-recursion during cleanup   Poul-Henning Kamp
• [master] 6e8e590 Set correct mode on -C mode workdir   Poul-Henning Kamp
• [master] 391d472 Have VCC report VMODs to Varnishd out of band (like VCL dependencies).   Poul-Henning Kamp
• [master] dfe4b5e Create vmod_cache subdirectory in master rather than child   Poul-Henning Kamp
• [master] 4752b12 Move the VMOD caching from worker to master to avoid too many Jail contortions.   Poul-Henning Kamp
• [master] 5c46217 Check -T and -M arguments sooner.   Poul-Henning Kamp
• [master] 2986c1f Ok, not quite _that_ early, we need to have access to SHM.   Poul-Henning Kamp
• [master] ab8c1f5 Typo   Poul-Henning Kamp
• [master] 3b30c88 Fix an off-by-one error in the place-holder VSLbin()   Poul-Henning Kamp
• [master] 4060412 Deal with WINDOWS_UPDATES on closed streams   Poul-Henning Kamp
• [master] 550b926 Hold lock over h2req->vsl   Poul-Henning Kamp
• [master] 9029079 Improve test coverage   Poul-Henning Kamp
• [master] e5a7775 More coverage   Poul-Henning Kamp
• [master] 429d2a1 Log all received frames, even if we throw them away.   Poul-Henning Kamp
• [master] b335906 Make H2_F_* a self-describing object and ditch the enum.   Poul-Henning Kamp
• [master] d8cef06 Add transmit properties to the frame descriptors.   Poul-Henning Kamp
• [master] a1eb3da Due to the incompetent H2 protocol design, we also need to keep track of which per-frame flags only get sent on the last frame of a continued frame-sequence.   Poul-Henning Kamp
• [master] e1e2c72 Overhaul H2 Settings handling   Poul-Henning Kamp
• [master] 1468655 Missed these in previous commit.   Poul-Henning Kamp
• [master] 7b13cad Don't use symbolic names for flags, GCC and Sun's C-compiler are not as smart as LLVM about const.   Poul-Henning Kamp
• [master] 3adc381 Make this test jail-compatible   Poul-Henning Kamp
• [master] ed9d11c More test coverage   Poul-Henning Kamp
• [master] 917a2c3 Ok, drop that test, it's not portable   Poul-Henning Kamp
• [master] 414d9f7 Add "-hdrlen <header> <length>"   Poul-Henning Kamp
• [master] 39211f9 Add a debug option for setting bogus H2 settings so VTC's dont have to test with 16k frame sizes.   Poul-Henning Kamp
• [master] b8506b3 Test H2 Continuation headers   Poul-Henning Kamp
• [master] c14a818 Move H2 Rx buf onto srq->ws   Poul-Henning Kamp
• [master] 856d562 The beginnings of a TX-scheduling facility.   Poul-Henning Kamp
• [master] 2087b3c Simplify the H2:body delivery   Poul-Henning Kamp
• [master] e32a440 Synchronize harder with the VSL thread in vtc_varnish.c   Poul-Henning Kamp
• [master] d9698da Streamline the H1->H2 handover a bit   Poul-Henning Kamp
• [master] 1a2a70f Add text about -I   Poul-Henning Kamp
• [master] 64d2b7c More fodder for release notes   Poul-Henning Kamp
• [master] 025011a Make vrt.h standalone again, and add test to ensure it stays that way.   Poul-Henning Kamp
• [master] e706b22 Cargo-culting for glorious victory for autocrap hegemony   Poul-Henning Kamp
• [master] f5179ab Try to isolate this even better.   Poul-Henning Kamp
• [master] 3db2f6a Constify   Poul-Henning Kamp
• [master] 39ea60c Some work on H2 RST_STREAM and GOAWAY   Poul-Henning Kamp
• [master] 0bc64c6 A bit more H2 polishing in the train.   Poul-Henning Kamp
• [master] 8cecc14 Simplify   Poul-Henning Kamp
• [master] d430063 Simplify   Poul-Henning Kamp
• [master] 28f4e1f Tweak a couple of -l options and drop an needless -s option   Poul-Henning Kamp
• [master] ae60839 Split this test in two, A9 was getting too many C-compiler invocations for the gcov vtester to finish in 60 seconds.   Poul-Henning Kamp
• [master] f9e4d9c Tighten up the Session pulldown sequence and let GOWAY do it.   Poul-Henning Kamp
• [master] c011ffb VDP_FINI is not allowed to return failure.   Poul-Henning Kamp
• [master] 255b99a Add a new Req State to handle Expect & Connection, use it for H2.   Poul-Henning Kamp
• [master] 9a267ba Have HTTP1 use the new R_STP_TRANSPORT as well.   Poul-Henning Kamp
• [master] 9d51cf7 Make H2 handle the waiting list.   Poul-Henning Kamp
• [master] 055f33d Start the worker (after -I processing) if we have a VCL and are not in -d mode.   Poul-Henning Kamp
• [master] 8649ed1 Add HTTP/2 status, minor tweaks.   Poul-Henning Kamp
• [master] fd186d9 Fix conditional requests in H2.   Poul-Henning Kamp
• [master] 18e425b Keep the H2 rx window open.   Poul-Henning Kamp
• [master] 2f40990 Staticize   Poul-Henning Kamp
• [master] 4042ed2 Use the right h2req for sending window updates   Poul-Henning Kamp
• [master] e3ef3cf Make sure the VSM overflow code has at least been executed once.   Poul-Henning Kamp
• [master] 7abd9d3 Try to make these two tests more predictable.   Poul-Henning Kamp
• [master] 7fd78ac Add a VARNISHTEST_DURATION environment variable to allow setting the test timeout (-t) through a layer of makefiles (aka: make check).   Poul-Henning Kamp
• [master] d431018 Start saying Via 5.1   Poul-Henning Kamp
• [master] a806b0c Final H2 bugfix before 5.1: Don't accidentally rip the session out because of a H1 assumption in transport-independent code.   Poul-Henning Kamp
• [master] fca21c7 Fix bug in H1 session cleanup.   Poul-Henning Kamp
• [master] 7b70617 One more H1 session cleanup fix and a lot more asserts.   Poul-Henning Kamp
• [master] b40e69b Explain why 5.1.1 is necessary   Poul-Henning Kamp
• [master] d87dcf1 This assert will not hold in all error situations.   Poul-Henning Kamp
• [master] 29db5c9 Add a H1 testcase which exercises the single like of code which triggers this bug for H2. (H2 will be added to the testcase as part of the fix.)   Poul-Henning Kamp
• [master] 5fd8055 Add a xport->req_fail() for H2   Poul-Henning Kamp
• [master] 3da40ca Add panic function for {vrt|vmod}_privs   Poul-Henning Kamp
• [master] e94a5e4 Also release the req0.   Poul-Henning Kamp
• [master] 4204414 Move a bunch of SES_ functions to cache_transport.h, they are not for casual calling.   Poul-Henning Kamp
• [master] 5b70451 Fixes to and better test-coverage of exception paths in H2 upgrade.   Poul-Henning Kamp
• [master] 4d3037a Generalize HPACK length encoding   Poul-Henning Kamp
• [master] 569dc01 Add a couple of commits to try to catch #2257   Poul-Henning Kamp
• [master] 8e77985 If the client workspace is almost, but not quite exhaused, we may not be able to get enough iovec's to do Chunked transmission.   Poul-Henning Kamp
• [master] 3bb4641 FlexeLinting: Avoid cpp macro argument names inside strings   Poul-Henning Kamp
• [master] d73a4e4 Flexelinting: initialize variable flexelint cannot see through.   Poul-Henning Kamp
• [master] 30786d6 Tune for 32 bit machines also   Poul-Henning Kamp
• [master] 1ed2adc Merge branch 'master' of github.com:varnishcache/varnish-cache   Poul-Henning Kamp
• [master] ce9c09a Negative SIZE to debug.workspace_allocate() now means "leave this much free in workspace".   Poul-Henning Kamp
• [master] 3ca71eb Add testcase for #2275   Poul-Henning Kamp
• [master] 01d7080 Use the new "allocate all but" debug.workspace_allocate() feature   Poul-Henning Kamp
• [master] 97d8ef9 Change/simplify the V1L logic for chunked encoding.   Poul-Henning Kamp
• [master] b335c0e Put a priority block in a HEADERS frame   Poul-Henning Kamp
• [master] 2e5da22 Make -vsl_catchup bail on test-failure or child death   Poul-Henning Kamp
• [master] 6358f4a Try to push H2 into even weirder corners.   Poul-Henning Kamp
• [master] cc5c71b Handle RST_STREAM on completed streams   Poul-Henning Kamp
• [master] e8bb7ec Give this test 8 bytes more to succeed in.   Poul-Henning Kamp
• [master] 35d1e51 Avoid sign-extension in hexdump   Poul-Henning Kamp
• [master] 1ad7006 A LOT more H2 plumbing.   Poul-Henning Kamp
• [master] 2aff3e5 Slightly more coverage.   Poul-Henning Kamp
• [master] 581948c Log the remote address to make it easier to cross-ref to VSL   Poul-Henning Kamp
• [master] 15f5c12 When the client abandons a fetch we close the connection to the backend if still in mid-transfer. With sufficiently small -bodylen, as in this test, that might or might not happen.   Poul-Henning Kamp
• [master] c022cc4 It is more robust to wait for the client connection to close than to delay a random-ish number of seconds   Poul-Henning Kamp
• [master] 54b5a09 Try to confuse GCC to not doing FP brain-damage.   Poul-Henning Kamp
• [master] 89e4f96 Remove place_holder code.   Poul-Henning Kamp
• [master] 735b0ec Add a copy of the rxthread to h2_sess and create an assert so we can mark up functions where only it can go.   Poul-Henning Kamp
• [master] ec9c04d Stop nagging the VSM if we ran out of it after an error   Poul-Henning Kamp
• [master] bf6ce0f Improve the handling of h2 frame-reception when the connection closes. Insist we get the frames we expect.   Poul-Henning Kamp
• [master] d56a248 Use each streams workers condvar instead of broadcasting.   Poul-Henning Kamp
• [master] 49c58d7 Don't let stream threads muck around in the stream list.   Poul-Henning Kamp
• [master] ad75b0e Add an assert so silly buggers who are to busy to check the arguments Lck_CondWait() will stopped in the tracks.   Poul-Henning Kamp
• [master] 9042625 This silly bugger has now checked the arguments to Lck_CondWait()   Poul-Henning Kamp
• [master] c737577 Straighten out the HPACK error path.   Poul-Henning Kamp
• [master] 3db481a A *little* bit more coverage   Poul-Henning Kamp
• [master] 8f33a90 These are fatal as well   Federico G. Schwindt
• [master] 994f073 Polish   Federico G. Schwindt
• [master] ae15fd7 Don't increment the hitpass stat for HfM   Federico G. Schwindt
• [master] a0249cf Merge CNUM and '-' handling   Federico G. Schwindt
• [master] 75da9a8 Add an HFP record example   Federico G. Schwindt
• [master] 0144634 This test requires 64bits   Federico G. Schwindt
• [master] 409114f Tidy up   Federico G. Schwindt
• [master] 520442b This never really worked so not a change per se   Federico G. Schwindt
• [master] 7841984 Spelling   Federico G. Schwindt
• [master] f4faa6e Fold r00936.vtc into vcc_action.c tests   Federico G. Schwindt
• [master] 0ce6f9b Fix build on OSX   Federico G. Schwindt
• [master] 6c4e164 First stab at osx on travis-ci   Federico G. Schwindt
• [master] 8ac8115 Switch to more recent linux and osx versions   Federico G. Schwindt
• [master] a14e514 Cosmetic change to kick travis-ci   Federico G. Schwindt
• [master] 4abdd29 Add these back - they are needed   Federico G. Schwindt
• [master] f4bd388 Switch approach while I speak with the travis guys   Federico G. Schwindt
• [master] ffa620d bah, trusty does not have nghttp2   Federico G. Schwindt
• [master] dcedbd2 Try to fix the annoying osx setup   Federico G. Schwindt
• [master] cc2af9e When using -C remove temporary directories on exit   Federico G. Schwindt
• [master] 2adcd88 Update   Federico G. Schwindt
• [master] ee5b896 Update config as per travis' support suggestion   Federico G. Schwindt
• [master] bd4a8ca Back to trunk   Federico G. Schwindt
• [master] c8ec10c Whitespace OCD   Federico G. Schwindt
• [master] 650b78b Update docs for *.uncacheable in generate.py regarding hit-for-miss/-pass   Geoff Simmons
• [master] f464a94 Revive use of the VSL HitPass tag.   Geoff Simmons
• [master] 70923dc Add SLT tag HitMiss to log hits on hit-for-miss objects.   Geoff Simmons
• [master] a95f496 Document the use of HitMiss and HitPass in VSL in What's New for 5.1.   Geoff Simmons
• [master] 2843de8 Added a placeholder chapter about HTTP/2 support to What's New for 5.1.   Geoff Simmons
• [master] e274785 Add the "Issues fixed" list to changes.rst for 5.1. This is just what I could find with "Fixes" or "Closes" in git log varnish-5.0.0..HEAD   Geoff Simmons
• [master] d906981 The test for standalone vrt.h runs on compilers that insist on a .c source.   Geoff Simmons
• [master] 5f267a4 Restructure the chapter headings in the performance doc in the User's Guide. It had subsections of the Cookies chapter that had nothing to do with them.   Geoff Simmons
• [master] 8cc11c1 Add chapters about cache misses and uncacheable content to the User's Guide. Primarily to document the varieties of pass (pass, hit-for-miss and hit-for-pass). Reviewed by Dridi.   Geoff Simmons
• [master] 37de1a0 Document the use of req.hash_always_miss to end a hit-for-pass state.   Geoff Simmons
• [master] aac0281 Start filling out "Upgrading to Varnish 5.1".   Geoff Simmons
• [master] 5e8953f Some minor additions to Upgrading to Varnish 5.1   Geoff Simmons
• [master] 6dbe82e Typos   Geoff Simmons
• [master] 1bc2479 Update the OPTIONS section in varnishd(3), and add a section about CLI files.   Geoff Simmons
• [master] 5679cc7 Summarize changes in varnishd command-line options in Upgrading to 5.1   Geoff Simmons
• [master] 9379ac9 Correct the notes on bereq. and beresp.backend in Upgrading to 5.1.   Geoff Simmons
• [master] 4b2db5c Added some miscellaneous notes to Upgrading to 5.1.   Geoff Simmons
• [master] e8f004d Summarize the interpretation of + and - with the various data types in "Upgrading to 5.1".   Geoff Simmons
• [master] 8f3ce04 Fix a double double word word.   Geoff Simmons
• [master] 65c6290 Add miscellaneous notes to Upgrading to Varnish 5.1.   Geoff Simmons
• [master] 91047a9 Grammar OCD   Geoff Simmons
• [master] b72f755 Add some bugfixes to changes.rst.   Geoff Simmons
• [master] a009f79 Update changes.rst for version 5.1.   Geoff Simmons
• [master] 1b80bb8 Document "math first, then string conversion"   Geoff Simmons
• [master] 6324d39 It's resp.reason, not req.reason   Geoff Simmons
• [master] cfce68d Grammar   Geoff Simmons
• [master] 444a622 Document that params vcl_dir and vmod_dir are deprecated.   Geoff Simmons
• varnish-commit：致所有HR的一封信 9:40:11   eiyo
• varnish-commit:合伙人的估值模型 12:35:50   fyrqopkma
• varnish-commit：您想学习阿里创新管理方面的成功经验吗？ 32532   oca
• 转发：向上管理的9个内容和5个步骤   varnish-commit
• varnish-commit--学会规避合伙人风险的4种方法 4:05:58   xnykgeps

Last message date: Fri Mar 31 17:14:06 CEST 2017
Archived on: Fri Mar 31 17:14:07 CEST 2017

• Messages sorted by: [ thread ] [ subject ] [ date ]
• More info on this list...