rembrant777 / db-engine Goto Github PK

add pthread's useful api's unit test case with name of pthread_test.cpp includes

phtread_t, pthread_attr_t, pthread_mutex_t, pthread_cond_t;

pthread_create
pthread_mutex_init
pthread_mutex_destroy
pthread_mutex_lock
pthread_mutext_unlock
pthread_cond_init
pthread_cond_destroy
pthread_cond_signal
pthread_cond_broadcast
pthread_join
pthread_exit

emeralddb is originally designed and implemednted based on the hash map.
what we need to enhance and explore is design uniform api that can shielding underlying details supports
multiple structure like:
B tree, B+ tree Bw and lsm (not sure for now just add a todo) and so on, uniform api deisgn also provides convenience to benchmark testing cases' execution .

Support actions to compile project source and execute unit test cases.

add bson as 3rd library to current project and implement CMakeLists.txt to support bson compile by cmake

try to support tree lsm structures to replace original hash table structure this is the first step
and in second step try to implement those structures based on rdma devices and based on the operations like wqe that rdma device support

Recently we got a failure in GitHub merge workflow after enabling in cmake file automatically quietly install gtest.

failure link address: https://github.com/Rembrant777/db-engine/actions/runs/4275539172

It is not maintainable to set gtest installation logic in cmake's CMakeLists.txt, cuz local env already have gtest installed, everytime execute local cmake will duplicate download. And in github workflow download not stable and difficult to maintain.

Move the gtest download && installation as a step same as boost's package download, compile and installation.

in this issue we add codes for client module, all codes locate in client/
branch name: issue_client

todo:
we also need add a system memory leaking tools in the github workflow for detecting whether each push codes may cause memory issues.

1. include google protobuf's closure as method pointer for thread pool
2. add install protobuf in github work flow step

Add gtest framework as 3rd library to support method and module unit test

for now, we support boost and gtest install automatically in github workflow everytime user executes pull request or push operations.

it is ok for gtest install and linking time, however, boost's package download, compile and install takes a too long time.

there are two ways can solve this time-consuming step in the workflow which are worth investigating:

1. download binary(compiled) file directly and move the header files to /usr/local/include && libs to /usr/local/libs directly -- question: file to large and there may exists an in-compatible from the binary files to current system
2. set boost cache refer to this github workflow file demo.

in previous db-engine project compiling and packaging all depend on multiple complex steps which often met linux system problems(or even cannot support mac local compiling).
like if we need to compile a project often follow those steps

autoscan  
# autoscan will generte a intermidiate file with the name of configure.scan 
# we need to rename the file from configure.scan to configure.ac 
aclocal 

autoconf 
autoheader 
automake --add-missing # this steps often need meet lots of errors 
./configure CXXFLAGS= CFLAGS= 
# after we execute the configure we often got a Makefile 
# finally execute make operaiton 
make 

in order to reduce those kind of issues happen and improve coding efficienty. I would like to choose cmake/CMakeLists.txt this compiling tool to liberate me from previous autoscan, automake those complex tools. In this way what I care about is how to write CMakeLists.txt and how to organize the parameters/flags that need to passing to the project.

Add a todo list:
branch: us_action

1. stage-1: execute cmake and pass is ok
2. stage-2: support automation test cases execution based on github's actions workflow
3. stage-3: this workflow should support code style, grammar checking, test case, regression and integrations like benchmark automaticatlly invoked by pull requests and merge operations

instead of sending data that stored in a buffer pointed by the char* we need to design the message body(struct) that contains message self detail metadata, message header and message body. also message type should be varied from different transporting scenarios.

add class to encapsulate thread pool basic operations

todo:
we also need to add a step or a new job for checking memory-leaking issues with the help of google sanitizers.

here is a demo for google sanitizer usage sample

Support network principle communication module(socket) and corresponding test cases.

branch: us_network

Add more logs support info, debug, error and trace

design docs are required(test cases. and modules)

https://toroidal-code.github.io/cppspec/

for now we already support gtest ut cases automatically executed every time we pull a request, what's more we also require an extra step that gonna generate test case coverage report and memory leak analyzing report just after unit test execution step.