| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
| |
writing to zen and horde upstreams. Return failure from HandleGetCacheRecord if the requested type does not match the cachetype. (#55)
* Fix bug with getting values PUT to Jupiter as CompressedBinary. When getting CompressedBinary records from Jupiter, they are expected to now be a record with a reference to the compact binary. This has to be accounted for when performing upstream GETs.
* HandleGetCacheRecord: avoid crashing on invalid type, and avoid sending back data that doesn't match the AcceptType.
|
| | |
|
| |
|
|
| |
This change ensures we retain the right content type of kCompressedBinary when propagating values from Zen->UpstreamZen. This is done via an RPC that posts a CbPackage. Furthermore when propagating from Zen->Jupiter, it composes its own referencing CbObject for them instead of sending a octet content type and Jupiter defining the referencing CbObject. When fetching Values from Jupiter, this new composed CbObject is still interpreted correctly by Zen.
|
| | |
|
| |
|
|
| |
ValueContentId where its a hash instead of an oid.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
| |
are used without an explicit definition
Also fixed up various code to compile with this, by using ZEN_THIRD_PARTY_INCLUDES_START/ZEN_THIRD_PARTY_INCLUDES_END macros
Removed prewindows.h/postwindows.h since they are no longer to be used due to the above
|
| | |
|
| |
|
|
|
|
|
|
| |
BinaryWriter/BinaryReader now implements memory buffer functionality which previously needed two chained instances of a Buffer/Reader. This was originally expected to be an abstraction for file and other stream access but this is not going to be useful so may as well collapse the functionality.
This also eliminates the need for stack-aware ref-counting which is the real reason for wanting to get rid of this code. This was a very old experimental feature which turned out to be a bad idea.
This also removes the /cas/batch endpoint
|
| | |
|
| |
|
|
| |
health/reconnecting at regular intervals.
|
| |
|
|
| |
Also removed snapshot_manifest (remnants of vfs prototype)
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
| |
* Code should no longer directly `#include spdlog/spdlog.h`, instead use `#include <zencore/logging.h>`
* Instead of explicit calls to `spdlog::info(...)` and such please use the logging macros defined in `zencore/logging.h`. I.e `ZEN_INFO`, `ZEN_DEBUG`, `ZEN_TRACE`, `ZEN_ERROR`, `ZEN_CRITITCAL`
* The macros will pick up the "most local" logger via a `Log()` call to retrieve a logger instance. To override the default logger in a class please implement your own `Log()` function
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |\ |
|
| | | |
|
| |/ |
|
| |
|
|
| |
Also implemented basic upstream query interface, which needs a bit more work to be fully functional (chunk propagation / fetching and new propagation policies as per DDC requirements)
|
| | |
|
| |
|