Logo, favicons and community files

.github/ISSUE_TEMPLATE/bug_report.yml

@@ -0,0 +1,42 @@
+name: Bug report
+description: Something behaves differently than it should
+labels: ["bug"]
+body:
+  - type: textarea
+    id: what
+    attributes:
+      label: What happened
+      description: What you saw, and what you expected instead.
+    validations:
+      required: true
+  - type: textarea
+    id: reproduce
+    attributes:
+      label: How to reproduce
+      description: >
+        The smallest sequence of statements or client calls that triggers it.
+        A failing test is ideal.
+      render: sql
+    validations:
+      required: true
+  - type: input
+    id: version
+    attributes:
+      label: Version or commit
+      placeholder: 0.2.0, or a commit hash
+    validations:
+      required: true
+  - type: input
+    id: platform
+    attributes:
+      label: Platform
+      placeholder: Linux x86_64, Windows 11, arm64 container, and so on
+    validations:
+      required: true
+  - type: textarea
+    id: extra
+    attributes:
+      label: Anything else
+      description: >
+        Server logs, whether it survives a restart, whether it reproduces on a
+        fresh data directory.

.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1,8 @@
+blank_issues_enabled: true
+contact_links:
+  - name: Security vulnerability
+    url: https://github.com/martian56/quantadb/security/policy
+    about: Do not open a public issue. Report it privately instead.
+  - name: Benchmark methodology
+    url: https://github.com/martian56/quantadb/blob/main/docs/benchmarks.md
+    about: How performance is measured here, and why single runs are not trusted.

.github/ISSUE_TEMPLATE/feature_request.yml

@@ -0,0 +1,22 @@
+name: Feature or SQL support
+description: Something QuantaDB cannot do yet
+labels: ["enhancement"]
+body:
+  - type: textarea
+    id: need
+    attributes:
+      label: What you are trying to do
+      description: >
+        The problem, rather than the feature. It often turns out there is a
+        shorter way to solve it.
+    validations:
+      required: true
+  - type: textarea
+    id: shape
+    attributes:
+      label: What it might look like
+      description: The SQL, the API, or the protocol behaviour you have in mind.
+  - type: textarea
+    id: alternatives
+    attributes:
+      label: How you work around it today

.github/ISSUE_TEMPLATE/performance.yml

@@ -0,0 +1,49 @@
+name: Performance
+description: Something is slower than it should be
+labels: ["performance"]
+body:
+  - type: markdown
+    attributes:
+      value: >
+        Performance work here is decided by measurement, so this template asks
+        for numbers. A report without them is still welcome, but it will start
+        by someone trying to reproduce what you saw.
+  - type: textarea
+    id: observed
+    attributes:
+      label: What you measured
+      description: >
+        Throughput or latency, the workload that produced it, and what you
+        expected instead.
+    validations:
+      required: true
+  - type: textarea
+    id: how
+    attributes:
+      label: How you measured it
+      description: >
+        The client, the concurrency, the row count, and how long each run
+        lasted. If you used loadgen or the field driver, the exact flags.
+    validations:
+      required: true
+  - type: input
+    id: hardware
+    attributes:
+      label: Hardware and storage
+      description: >
+        CPU count, memory, and what the data directory sits on. Flush cost
+        varies by orders of magnitude between an SSD, a spinning disk and a
+        container volume, and it usually explains the result.
+      placeholder: 4 vCPU, 8 GiB, NVMe SSD, ext4
+    validations:
+      required: true
+  - type: dropdown
+    id: repeated
+    attributes:
+      label: Did you run it more than once?
+      options:
+        - "Yes, and the runs agreed"
+        - "Yes, and the runs disagreed"
+        - "No, one run"
+    validations:
+      required: true

.github/pull_request_template.md

@@ -0,0 +1,28 @@
+<!--
+Performance claims are settled by the bench comment below, not by local
+timings. A change under about fifteen percent needs a second run to mean
+anything.
+-->
+
+## What this changes
+
+<!-- One paragraph. What was wrong, and what it does instead. -->
+
+## Why
+
+<!-- Closes #NNN, or the reasoning if there is no issue. -->
+
+## How it was checked
+
+- [ ] `cargo test --workspace`
+- [ ] `cargo clippy --workspace --all-targets -- -D warnings`
+- [ ] `cargo fmt --all`
+- [ ] Tests added for the behaviour that changed
+
+## Anything a reviewer should look at closely
+
+<!--
+Durability and concurrency changes are the ones that hurt later. If this
+touches the commit path, recovery, or page reuse, say what ordering it
+relies on.
+-->

CODE_OF_CONDUCT.md

@@ -0,0 +1,39 @@
+# Code of conduct
+
+## The short version
+
+Be decent to people. Argue about code, not about each other.
+
+## What is expected
+
+- Assume the person you are replying to is competent and acting in good faith.
+- Criticise the design, the benchmark, or the patch. Not the person who wrote it.
+- Accept that other people have different priorities, hardware, and workloads.
+- When you are wrong, say so plainly and move on. It costs nothing here.
+
+## What is not acceptable
+
+- Personal attacks, insults, or demeaning comments.
+- Harassment of any kind, public or private, including unwelcome attention.
+- Publishing anyone's private information without their explicit permission.
+- Sustained disruption of discussions, issues, or reviews.
+
+## Scope
+
+This applies in the issue tracker, pull requests, discussions, and anywhere
+someone is representing the project.
+
+## Reporting
+
+Report anything that needs attention to fuadelizade6@gmail.com. Reports are
+read by the maintainer and are not shared further without your agreement.
+
+Expect an acknowledgement within a few days. Outcomes range from a private
+word to a permanent ban, depending on what happened. The maintainer will not
+act on a report while personally involved in the dispute without asking a
+neutral third party to look first.
+
+## Attribution
+
+This is adapted in spirit from the [Contributor Covenant](https://www.contributor-covenant.org),
+rewritten to be shorter and to say what it actually means.

CONTRIBUTING.md

@@ -0,0 +1,71 @@
+# Contributing
+
+Thanks for looking. QuantaDB is a young database and the useful contributions
+right now are small, measured, and reviewable.
+
+## Before you write code
+
+Open an issue first for anything beyond a fix. It saves you building something
+that does not fit, and it saves a review that has to say no.
+
+If you are looking for somewhere to start, the issue tracker carries the
+current performance work with its measurements attached, which is usually the
+clearest statement of what the project needs.
+
+## Getting set up
+
+Rust 1.96 or newer.
+
+```bash
+cargo build --workspace
+cargo test --workspace
+cargo clippy --workspace --all-targets -- -D warnings
+cargo fmt --all
+```
+
+All four have to be clean; CI runs them as separate jobs so a formatting slip
+does not hide behind a compile.
+
+## What a good change looks like
+
+**One issue per pull request**, with several small commits rather than one
+large one. Each commit should build and pass tests on its own.
+
+**Tests that would have caught the bug.** For storage and MVCC work that
+usually means a test that survives a reopen, or one that drives the exact
+interleaving that broke.
+
+**Comments only where the code cannot speak.** Keep them short. Explain why
+something is the way it is, not what the next line does.
+
+**Commit messages that say what changed and why.** A subject line under about
+seventy characters, a blank line, then prose. No trailing metadata.
+
+## Performance changes
+
+This is the part that is different here, and it is not negotiable.
+
+**A performance claim needs numbers from the bench matrix, not from your
+laptop.** Every pull request is benched on five machine profiles against the
+previous build running on the same runner, and the result is posted as a
+comment. Local timings have repeatedly pointed the opposite way to the truth
+on this project, so they do not decide anything.
+
+**One run is not evidence.** Runner speed varies. A change under about fifteen
+percent needs a second run before anyone believes it, and the report says so.
+
+**Negative results are worth submitting.** Several issues here exist only to
+record what did not work and why, so nobody spends the idea twice. If you tried
+something reasonable and the matrix rejected it, that is a useful pull request
+description even if the code never merges.
+
+## Licensing
+
+QuantaDB is Apache-2.0. By opening a pull request you agree that your
+contribution is licensed under the same terms, and you confirm you have the
+right to contribute it. Commits should carry a `Signed-off-by` line asserting
+the [Developer Certificate of Origin](https://developercertificate.org):
+
+```bash
+git commit -s
+```