7.1 #8
7.1 #8
Conversation
to work with 2.71
Vmods are banned from cache_param for quite some time now, so the vmod should have its own parameter for the gzip buffer size.
Much more can be done here to reduce the required varnish internal headers, but for now this commit concludes the required changes for varnish-cache master as of now (post 7.1)
varnish-cache serializes the vmod events, so, unless the vmod itself has any concurent access, no mutex is required.
The PR to bring it up to speed is not yet merged, but I hope that will change soon Refinitiv/libvmod-soap#8
|
|
||
| #define POOL_KEY "VRN_IH_PK" | ||
|
|
||
| static pthread_mutex_t soap_mutex = PTHREAD_MUTEX_INITIALIZER; |
There was a problem hiding this comment.
what's the rationale behind this ?
There was a problem hiding this comment.
see 47ccd3e
varnish-cache serializes the vmod events, so, unless the vmod itself has
any concurent access, no mutex is required.
| { | ||
| struct priv_soap_task *soap_task; | ||
|
|
||
| AN(priv); |
There was a problem hiding this comment.
shall we preserve the AN(priv) ?
There was a problem hiding this comment.
CAST_OBJ_NOTNULL() two lines down already asserts on the value not being NULL
Some changes are incompatible in varnish-cache. There are some vmods which try to retain backwards compatibility using macros, wrappers etc. but I would recommend against this practice and just branch off the current code as, say, a |
This series of changes makes the vmod work with varnish-cache post-7.1 (master as of varnishcache/varnish-cache@142c3a0)
Much more could be done, consider this PR as a first "time to market" patch series
Also includes #7 which might be justified should this PR not be accepted