Serializer requires class to be defined, so it has to be in .h file. It
also does not support nested types yet, so move it outside of containing
class.
python3 needs to install pyparsing excplicitely. This adds the
installation of python3-pyparsing to the require dependencies in the
README.md
Signed-off-by: Amnon Heiman <amnon@scylladb.com>
the serializer
This patch adds a specific template initialization so that the rpc would
use the serializer and deserializer that are auto-generated.
Signed-off-by: Amnon Heiman <amnon@scylladb.com>
This patch adds the serializer and serializer_imple files. They holds
the functions that are not auto generated: primitives and templates (map
and vector) It also holds the include to the auto-generated code.
Signed-off-by: Amnon Heiman <amnon@scylladb.com>
This patch adds rules and the idl schema to configure, which will call
the code generation to create the serialization and deserialization
functions.
There is also a rule to create the header file that include the auto
generated header files.
Signed-off-by: Amnon Heiman <amnon@scylladb.com>
This is a definition example for gossip_digest_ack with all its sub
classes.
It can be used by the code generator to create the serializer and
deserializer functions.
Signed-off-by: Amnon Heiman <amnon@scylladb.com>
inet_address uses uint32_t to store the ip address, but its constructor
is int32_t.
So this patch adds such a constructor.
Signed-off-by: Amnon Heiman <amnon@scylladb.com>
This patch contains two changes, it make the constructor with parameters
public. And it removes the dependency in messaging_service.hh from the
header file by moving some of the code to the .cc file.
Signed-off-by: Amnon Heiman <amnon@scylladb.com>
The code generation takes a schema file and create two files from it,
one with a dist.hh extension containing the forward declarations and a second
with dist.impl.hh with the actual implementation.
Because the rpc uses templating for the input and output streams. The
generated functions are templates.
For each class, struct or enum, two functions are created:
serialize - that gets the output buffer as template parameter and
serialize the object to it. There must be a public way to get to each of
the parameters in the class (either a getter or the parameter should be
public)
deserialize - that gets an input buffer, and return the deserialize
object (and by reference the number of char it read).
To create the return object, the class must have a public constructor
with all of its parameters.
The solution description can be found here:
https://github.com/scylladb/scylla/wiki/Serializer-Deserializer-Code-generation
Signed-off-by: Amnon Heiman <amnon@scylladb.com>
Only difference from previous sleep is that we will
explicitly delete the objects if the process terminates
before tasks are run. I.e. make ASas happier.
Message-Id: <1453295521-29580-1-git-send-email-calle@scylladb.com>
Arguments buffer_size and true were accidently inverted.
GCC wasn't complaning because implicit conversion of bool to
int, and vice-versa, is valid.
However, this conversion is not very safe because we could
accidentaly invert parameters.
This should fix the last problem with sstable_test.
Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com>
Message-Id: <9478cd266006fdf8a7bd806f1c612ec9d1297c1f.1453301866.git.raphaelsc@scylladb.com>
test_cassandra_hash also sort of expects exceptions. ASas causes false
positives here as well with seastar::thread, so do it with normal cont.
Message-Id: <1453295521-29580-2-git-send-email-calle@scylladb.com>
Since 581271a243 "sstables: ignore data
belonging to dropped columns" we silently drop cells if there is no
column in the current schema that they belong to or their timestamp is
older than the column dropped_at value. Originally this check was
applied to row markers as well which caused them to be always dropped
since there is no column in the schema representing these markers.
This patch makes sure that the check whether colum is alive is performed
only if the cell is not a row marker.
Signed-off-by: Paweł Dziepak <pdziepak@scylladb.com>
Message-Id: <1453289300-28607-1-git-send-email-pdziepak@scylladb.com>
From Paweł:
These patches contain fixes for date and timestamp types:
- date and timestamp are considered compatible types
- date type is added to abstract_type::parse_type()
When this code was originally written, we used to operate on a generic
output_stream. We created a file output stream, and then moved it into
the generic object.
Many patches and reworks later, we now have a file_writer object, but
that pattern was never reworked.
So in a couple of places we have something like this:
f = file_object acquired by open_file_dma
auto out = file_writer(std::move(f), 4096);
auto w = make_shared<file_writer>(std::move(out));
The last statement is just totally redundant. make_shared can create
an object from its parameters without trouble, so we can just pass
the parameter list directly to it.
Signed-off-by: Glauber Costa <glauber@scylladb.com>
Message-Id: <c01801a1fdf37f8ea9a3e5c52cd424e35ba0a80d.1453219530.git.glauber@scylladb.com>
* Match origin log messages
- Demote per-file printouts to "debug" level.
* Print an all-files stat summary for whole replay (begin/summary)
- At info level, like origin
Prompted by dtest that expects origin log output.
Message-Id: <1453216558-18359-1-git-send-email-calle@scylladb.com>
"We need to be able to replay mutations created using older versions of
the table's schema. frozen_mutation can be only read using the version
it was serialized with, and there is no guarantee that the node will
know this version at the time of replay. Currently versions are kept
in-memory so a node forgets all past versions when it restarts. This
was not implemented yet, replay would fail with exception if the
version is unknown."
We need to be able to replay mutations created using older versions of
the table's schema. frozen_mutation can be only read using the version
it was serialized with, and there is no guarantee that the node will
know this version at the time of replay. Currently versions kept
in-memory so a node forgets all past versions when it restarts.
To solve this, let's store canonical_mutations which, like data in
sstables, can be read using any later schema version of given table.
From Paweł:
"This series contains some more fixes for issues related to alter table,
namely: incorrect parsing of collection information in comparator, missing
schema::_raw._collections in equality check, missing compatibility
information for utf8->blob, ascii->blob and ascii->utf8 casts."
