Sprint Backlog 02

Updates to sprint and product backlog.

With the move to ImGui the builds started failing across the board.

Disable other OSs for now.

We need to be able to represent the data in the filesystem for the ORE samples in a way that allows users to load and save the samples. The data in the filesystem has the following structure:

• Examples
  • Example 1:
    • Input
    • ExpectedOutput, e.g. Output
  • Example 2
  • …
  • Example n
  • Input

The overall context under which valuation is taking place needs a name. Options:

• environment
• workspace
• sandbox: seems to imply the data is not "real" but that is not always the case. This seems more like a label we could apply to the top-level container rather than its overall name.

Each entry within the top-level container is then itself another kind of container. Options:

• namespace
• package

There are two types of these containers:

• libraries: components whose content is made to be shared with other components; and
• executables: components whose content is sufficient to execute computations.
• component, elements.

Elements within a component is of two kinds: input and output. However, this may not necessarily be a good classification because some outputs can be used as inputs to output other elements. Perhaps this is more of a tag.

Notes:

• move this analysis to documentation.

We need a simple way to keep track of changes. No need for bitemporal, just single dimension.

Links:

• Track last modification timestamp of a row in Postgres

We need:

• latest
• as of
• all for a given iso code.

Notes:

• add timestamp fields to domain objects.

Links:

• Getting Started with ImGui: A Beginner’s Tutorial
• How to use libpqxx to receive notifications from the PostgreSQL database?
• Text color of disabled menu item is incorrect

We will forge ahead with a QT UI, given the issues we had with imGui cross platform. We can worry about it later.

To get started, download Gemini CLI 0.1.18 or later and run `/setup-github`.

Links:

• GH run-gemini-cli
• Meet your new AI coding teammate: Gemini CLI GitHub Actions

We never got it to compile under windows and osx.

• Reason: GRPC is quite convoluted and generates weird code. Try first to build a simple RPC using ZeroMQ and see how hard it is.

Needs to be able to login, and retrieve currencies.

Rationale: Actually we should experiment with raw asio first before adding more complexity.

Error:

misspell-fixer: -r Enable real run. Overwrite original files!
misspell-fixer: -s Enable showing of diffs.
misspell-fixer: -v Enable verbose mode.
misspell-fixer: -n Disable backups.
misspell-fixer: Target directories: .
misspell-fixer: Your grep version is +zstd which is at least the optimal: 2.28.
misspell-fixer: Iteration 0: prefiltering.
misspell-fixer: Results of prefiltering: (filename:line:pattern)
./doc/agile/v0/sprint_backlog_02.org:382:constraints
misspell-fixer: Iteration 0: processing.
misspell-fixer: actual file: ./doc/agile/v0/sprint_backlog_02.org
misspell-fixer: temp file: ./doc/agile/v0/sprint_backlog_02.org.64
misspell-fixer: misspellings are fixed!
misspell-fixer: Iteration 0: done.
misspell-fixer: Iteration 1: prefiltering.
misspell-fixer: Results of prefiltering: (filename:line:pattern)
misspell-fixer: Iteration 1: nothing to replace.
misspell-fixer: Iteration 1: done.
=================================

Error: Unable to process file command 'output' successfully.
Error: Invalid format '+++ ./doc/agile/v0/sprint_backlog_02.org.64	2025-09-20 10:57:51.112708427 +0000'

The problem may be related to the error in the URL.

• remove code using GRPC, leave only server and client skeletons.

Using the examples, create a really simple server and client. Must use SSL.

Links:

• cobalt docs
• GH cobalt

Links:

• GH reflect-cpp

Spend some time with AI to figure out an initial architecture for the system. Does not have to be perfect but must follow the data oriented principles we are trying to exercise.

Notes:

• each subsystem should have all related functionality:
  • risk: domain types for ORE, ORE adapters, DB support, messages, managers with rings, table dispatchers.
  • comms: basic asio and cobalt infrastructure. Frame, message sizes and its handling, socket handling, generic subsystem dispatcher. Frame needs to have a subsystem ID and a message ID. Basic asio server and client.
• client library contains a wrapper around client functionality, by subsystem. Each should have their own "client".
• server project contains the runnable service, at the start supporting all subsystems. In the future we can split these into multiple servers.
• console (cli) calls client library to exercise it.
• Consider naming server and client after the protocol (nexus).
• HTTP is implemented as a separate server.

Notes:

• it would be nice to be able to support SQLite as well. However, there is no common connection class, so we need to use templates.

Links:

• GH sqlgen
• #52: Support for temporal data

Rationale: We need to create the schema manually anyways.

We may need to update the SQL from sqlgen. This will make life easier.

As defined in analysis story, refactor CLI with new approach.

At present we are not able to see logs for tests.

We need to distinguish between the types in comms etc and the core types for domain modeling. We need a good name for this.

We need to see if we can parse ORE XML using reflect cpp instead of hand crafted code.

Links:

• std::variant, rfl::Variant and rfl::TaggedUnion

Issue was related to YAML spaces. GitHub copilot:

Analysis:

• The push trigger covers pushes to main and any tags.
• The problem is that pull_request is indented under push, making it a property of push (which is not valid YAML for GitHub Actions).
• In GitHub Actions, push and pull_request must be sibling keys under on:.
• Result: Your pull request builds are not being triggered on Windows, because the pull_request trigger is incorrectly nested.

At present we are getting random lisp snippets when filling in agile backlogs.

Does not solve all the problems but helps:

(defun cunene/org-mode-hook ()
  "Hook to run in org mode."
  ;; Make sure Org's native completions are prioritized
  (setq-local completion-at-point-functions
              (append (list #'org-pcomplete-initial) ; Org's native completion
                      completion-at-point-functions)))

Notes:

• fixed assorted ilist issues.
• fixed completion on git commit.

Links:

• (Bi)Temporal Tables, PostgreSQL and SQL Standard
• PostgreSQL 18 Temporal Constraints
• Waiting for PostgreSQL 18 – Add temporal FOREIGN KEY constraints
• PostgreSQL Identity Column
• SQL2011Temporal
• PostgreSQL 18 New Features
• Understanding Bitemporal Primary Keys
• Implementing System-Versioned Tables in Postgres
• Historical records with PostgreSQL, temporal tables and SQL:2011
• GH temporal_tables
• Query data in a system-versioned temporal table

As per the new architecture, we should have all the services within each subsystem.

Notes:

• merge tests into the project. Actually this screws up the include paths, for not a huge advantage. Cancelling it.
• clean up properties in domain entities.

Error: Target "ores.risk.lib" links to: Error(s) when configuring the project CMake Error at D:/a/OreStudio/OreStudio/CTest.cmake:276 (message): Error: Failed to configure

pugixml::static

but the target was not found. Possible reasons include:

• There is a typo in the target name.
• A find_package call is missing for an IMPORTED target.
• An ALIAS target is missing.

We are getting spurious memory leaks:

{
   Memcheck:Leak
   match-leak-kinds: reachable
   fun:malloc
   obj:/home/runner/work/_temp/-1583124222/cmake-4.1.1-linux-x86_64/bin/cmake
   obj:/home/runner/work/_temp/-1583124222/cmake-4.1.1-linux-x86_64/bin/cmake
   fun:__libc_start_main@@GLIBC_2.34
   obj:/home/runner/work/_temp/-1583124222/cmake-4.1.1-linux-x86_64/bin/cmake
   obj:*
   obj:*

Move comms related code to it's own library.

Tasks:

• add options for port, key file etc.
• add support for environment variables.

// Parse environment variables (prefix "MYAPP_")
po::store(po::parse_environment(desc, "MYAPP_"), vm);

• refactor server and client code into their own classes.

We need to setup the infrastructure for messaging within the risk subsystem.

Move all of the code in application related to repl into it's own class and tidy it up.

At present we are getting:

Additional packages (*) will be modified to complete this operation.
$VCPKG_BINARY_SOURCES: warning: The 'x-gha' binary caching backend has been removed. Consider using a NuGet-based binary caching provider instead, see extended documentation at https://learn.microsoft.com/vcpkg/users/binarycaching?WT.mc_id=vcpkg_inproduct_cli.
  on expression: clear;x-gha,readwrite

Raised:

• #253: Addressing vcpkg warning related to binary caching

It seems caching is not kicking in so builds are very slow.

After latest comms changes, the builds got borked.

We need to get latest vcpkg as we haven't updated in a while.

Notes:

• latest sqlgen seems borked.

Links:

• #70: Problem with queries and schemas after v0.3.0 update
• #69: Handling of boolean values on postgres
• #68: Deallocating prepared statements on error

Add the database tables, messages, commands etc to support logins.

Prompt:

In the project ores.accounts, add support for account management. It will include login and logout, password management, account locking and all the standard account management features. Implementation will be done in steps. Use the appropriate c++ 23 types, using boost where necessary to model the suggested fields. Add comments to all fields. Follow the exact same pattern as the risk project for both folder structure as well as the code.

First we need a new class in a domain folder called account with the key fields:

• int Version
• string ModifiedBy
• GUID Id
• string Username
• string PasswordHash
• string PasswordSalt
• string TotpSecret
• string Email
• bool IsAdmin

Then we need a class called logins with the following fields:

• GUID AccountId
• string LastIp
• string LastAttemptIp
• int FailedLogins
• bool Locked
• DateTime LastLogin
• bool Online

Once I am happy with the code you generate, I will give you instructions for the next steps.

I want to add a new class to accounts responsible for password management. Think of a good name. create a new folder for it, suitably named. It's job will be to create a password hash given a password, a hash and a salt using HMACSHA512. then add a verify password hash method which returns true if the password hashes correctly. Tell me what your implmementation plan is before you start.

problems:

• errors in create account request drop connection, should return an error.
• login insertion is not transactional, should rollback with account.
• ERROR: prepared statement "sqlgen_insert_into_table" already exists
• test email_unique_idx

Links:

• #68: Deallocating prepared statements on error

At present we are getting:

Building qtbase[brotli,concurrent,core,dbus,dnslookup,doubleconversion,egl,fontconfig,freetype,gui,harfbuzz,icu,jpeg,network,opengl,openssl,pcre2,png,sql,sql-psql,sql-sqlite,testlib,thread,widgets,xcb,xcb-xlib,xkb,xkbcommon-x11,xlib,xrender,zstd]:x64-linux@6.8.3#5...
CMake Warning at ports/qtbase/portfile.cmake:49 (message):
  qtbase currently requires packages from the system package manager.  They
  can be installed on Ubuntu systems via sudo apt-get install '^libxcb.*-dev'
  libx11-xcb-dev libglu1-mesa-dev libxrender-dev libxi-dev libxkbcommon-dev
  libxkbcommon-x11-dev libegl1-mesa-dev.

According to grok:

The CMake warning from ports/qtbase/portfile.cmake indicates that the qtbase package in vcpkg requires additional system dependencies (like libxcb and others) to be installed on your system, specifically for Ubuntu. This warning appears because vcpkg detects that these dependencies are not satisfied. To remove the warning, you need to install the required system packages or suppress the warning if you’re sure the dependencies are met or not needed.

We seem to be installing all of the required libraries on our script. We may need to:

export VCPKG_DISABLE_SYSTEM_PACKAGE_CHECK=1

The action to review PRs using gemini is failing.