The solution as a whole doesn't build yet - we probably want to remove
ProtoDump and ProtoMunge entirely, and ProtoBench should use Jan's new
benchmarks for parity with Java.
The version of NUnit on my machine, packaged with Mono 3.12.1, is
only NUnit 2.4.2, which is extremely old - it still requires an explicit
[TestFixture] attribute on test fixtures. I've added one just for ByteStringTest
for the moment so that we can see some tests passing in Travis, but as part of
a separate PR we should work on making sure we're using a recent NUnit version.
(It may already be doing so, but we can check that once it's working and merged.)
- Make some members internal
- Remove a lot of FrameworkPortability that isn't required
- Start adding documentation comments
- Remove some more group-based members
- Not passing in "the last tag read" into Read*Array, g
We'll probably want a lot of the code from the serialization project when we do JSON, but enough of it will change that it's not worth keeping in a broken state for now.
This is effectively reimplementing List<T>, but with a few advantages:
- We know that an empty repeated field is common, so don't allocate an array until we need to
- With direct access to the array, we can easily convert enum values to int without boxing
- We can relax the restrictions over what happens if the repeated field is modified while iterating, avoiding so much checking
This is somewhat risky, in that reimplementing a building block like this is *always* risky, but hey...
(The performance benefits are significant...)
This mirrors commit 7c86bbbc7a3365c034d82173b38ec4427b98b3b2 in the pull request to
the main protobuf project, but also reduces the size of the buffer created. (There's no point in
creating a 1024-byte buffer if we're only skipping 5 bytes...)
Remove ICodedInputStream and ICodedOutputStream, and rewrite CodedInputStream and CodedOutputStream to be specific to the binary format. If we want to support text-based formats, that can be a whole different serialization mechanism.
This makes repeated fields really awkward at the moment - but when we reimplement RepeatedField<T> to be backed by an array, we can cast the array directly...
Cache a reference to Encoding.UTF8 - the property access is (rather surprisingly) significant.
Additionally, when we detect that the string is all ASCII (due to the computed length in bytes being the length in characters), we can perform the encoding very efficiently ourselves.
We still have some protos which aren't generated how we want them to be:
- Until we have an option to specify the "umbrella" class, DescriptorProtoFile
will be broken. (The change of name here affects the reflection descriptor,
which accounts for most of the change. That's easier than trying to work out
exactly which occurrences of Descriptor need changing though.)
- That change affects UnittestCustomOptions
- Issue #307 breaks Unittest.cs
After this commit, we don't have the record of the fixups in the files themselves
any more, but one centralized record in the shell script.
To my surprise, executing generate_protos.sh used the version of Bash installed with Git for Windows by default.
After a few modifications to detect the most appropriate protoc to use, this worked pretty simply.
This change also:
- adds generation of the address book tutorial proto,
- fixes the addressbook.proto to specify proto2 explicitly (to avoid a warning from protoc; I don't think we want warnings...)
- fixes the addressbook.proto C# namespace (which I thought I'd done before, but apparently hadn't)
- includes the regenerated UnittestCustomOptions.cs apart from the DescriptorProtoFIle => Descriptor change
This is the start of establishing a C# namespace of "Google.ProtocolBuffers.TestProtos.Proto3" for proto3-syntax protos.
We could optionally split the directory structure as well into Proto2 and Proto3 for clarity.
