]> err.no Git - varnish/shortlog
varnish
2007-10-08 phkOnly attempt to decode status integer if we have a...
2007-10-08 phkAdd an assert for spotting ticket #162
2007-10-08 phkCatch EOF on reads instead of looping.
2007-10-08 phkDrop the asserts around shutdown(2), there are a number of
2007-10-08 phkupd
2007-10-08 phkMore acceptor work:
2007-10-08 phkBail normally instead of asserting on buffer full.
2007-10-08 phkUse HTC_Rx() directly instead of vca_pollsession()...
2007-10-08 phkcomments/style
2007-10-08 phkTry to make the pipe code even more robust
2007-10-08 phkFlush the threads shmlog before we try to make up a...
2007-10-05 ssmFix ticket #166, debian package issues
2007-10-01 phkGo over pipe's polling code and try to do it right...
2007-10-01 phkOops, last commit included some debugging
2007-10-01 phkRemember to move numeric status when we filter one...
2007-10-01 phkInitialize the obj->ws before we use it.
2007-10-01 phkMove the objects and their http header into storage...
2007-10-01 phkMove the workerthreads log from malloc to stack.
2007-10-01 phkEliminate a & which I tend to forget every single time.
2007-10-01 phkWe always want to go through vcl_fetch(), even on error 503
2007-09-30 phkAdd VCL action "esi" and a vrt file to contain it.
2007-09-30 phkMake the number of restarts a parameter.
2007-09-30 phkAdd the first stab at a "restart" mechanism.
2007-09-30 phkAdd req.restarts variable.
2007-09-30 phkAdd "restart" action for most VCL methods.
2007-09-30 phkEvict backend-health code which had strayed fra from...
2007-09-30 phkMore caching of HTTP status
2007-09-30 phkCache the numeric status code.
2007-09-30 phkSignificantly reduce level of paranoia.
2007-09-30 phkWe may not have a worker thread here, we could be calle...
2007-09-30 phkRemove unused #includes
2007-09-30 phknitpicking
2007-09-30 phkSet the poll'ed fd to -1 when we halfclose the fd to...
2007-09-29 phkAnother cleanup commit:
2007-09-28 phkComplete the split of HTTP-editing and HTTP-transmissio...
2007-09-28 phkInitial split of code which transfer HTTP protocol...
2007-09-28 phkMove the WS_ prototypes down where they belong.
2007-09-28 phkPush the http structure out of the early bits of the...
2007-09-28 phkPreparation for implementation of restarts: Move the...
2007-09-27 desUse CHECK_OBJ_ORNULL().
2007-09-26 phkdon't make acl's const, they are not.
2007-09-26 phkMore txt'ing
2007-09-26 phkanother place we can use Tlen
2007-09-26 phkUse the new txt type more extensively, it does improve...
2007-09-26 phkConvert the received request and pipelined bytes into...
2007-09-26 phkstruct http_hdr is more than that, it is a general...
2007-09-25 phkMore flexelint nit-picking of various sorts
2007-09-25 phkConstify a lot of the VRT API
2007-09-25 phkMega-Const'ification
2007-09-25 phkIf the C-compiler fails, we don't know if it created...
2007-09-25 phkConstification
2007-09-25 phkConstification of certain storage functions
2007-09-25 phkMake Tim_format() take a double time argument
2007-09-25 phkMake the arguments to CLI functions const-const.
2007-09-25 phkCatch up with sys/queue.h changes
2007-09-25 desThe previous commit made things worse rather than bette...
2007-09-25 desPortability: use -lsocket / -lns on systems that need...
2007-09-25 desUse a config cache.
2007-09-25 phkNew more paranoid settings for FlexeLint
2007-09-25 phkMake the logic visible for FlexeLint
2007-09-25 phkMisc changes:
2007-09-25 phkUnnecessary check
2007-09-25 phkBail if we get unknown address families
2007-09-25 phkAssert return values of system{calls,functions}
2007-09-25 phktypo
2007-09-25 phkPolish
2007-09-25 phkUse pdiff() to guard against negative point differences
2007-09-25 phkTwo consts is one const too many.
2007-09-25 phkMore const polishing
2007-09-25 phkMake id type "int"
2007-09-25 phkConst'ifcation
2007-09-25 phkPolish shmlog:
2007-09-25 phkWhen we take the difference between two pointers, the...
2007-09-25 phkUse the correct shmlog function.
2007-09-25 phkThe first of a long sequence of committs, inspired...
2007-09-24 cecilihfRe-enable the code causing bug #161 (which is now fixed)
2007-09-24 cecilihfFixes bug #161.
2007-09-24 phkActually, we cannot touch objects on deathrow, so just...
2007-09-24 phkRemove an assert which isn't valid: we could try to...
2007-09-24 phkMake the cli_timeout a parameter. The master will...
2007-09-24 cecilihfUpdate random and round-robin backends to use the new...
2007-09-24 phkAdd code to VCC to output an identifier for each simple...
2007-09-24 phkOur first cut at a LRU processing contained a three...
2007-09-24 phkAdd a couple of statistics counters for LRU
2007-09-24 phkFix a NULL pointer dereference which cannot currently...
2007-09-23 desReverse the logic in r1997 from !is_get to is_head.
2007-09-23 phkFix another cornercase that fell out as part of the...
2007-09-23 desUnbreak header ordering.
2007-09-23 phkWhen a workerthread dies of old age, make sure to nuke...
2007-09-23 phkPart of this commit sneaked into the previous #1985:
2007-09-23 phkPartially revert previous optimization: it wasn't.
2007-09-23 phkPolish, including use of CHECK_OBJ_ORNULL()
2007-09-23 phkUse CHECK_OBJ_ORNULL() macro
2007-09-23 phkAdd CHECK_OBJ_ORNULL() macro
2007-09-23 phkPolish while reading review.
2007-09-21 phkAdd an experimental runtime parameter for the lru_timeout.
2007-09-21 phkA temporary work-around for #161
2007-09-21 phkDon't leak 50 bytes when creating the new Date: header.
2007-09-21 phkDon't copy the Date: header from the object to the...
2007-09-21 phkMove implicit reference to first backend to the xref...
next