[metif12]

Summary

Adds a Gitea parity analysis, which the repository did not have anywhere. Gitly compares itself to GitHub and GitLab, but Gitea is the more useful bar: open source, single-node, same problem, same scale. Issue #142 asks for a Gitea row on the gitly.org comparison tab, and this is the factual basis anyone would need before writing one.

 README.md
-The current GitLab capability comparison ... is documented in [docs/gitlab-parity.md](docs/gitlab-parity.md).
+The current GitLab capability comparison ... is documented in [docs/gitlab-parity.md](docs/gitlab-parity.md).
+The comparison against Gitea, which is the closer-sized forge and therefore the more useful bar, is in [docs/gitea-parity.md](docs/gitea-parity.md).

 docs/gitea-parity.md  (new)

The source of truth is Gitea's own matrix, https://docs.gitea.com/installation/comparison, version 28. Gitea's caveat is carried into the document so readers know the bar moves:

"we don't regularly check for feature changes in other products, so this list may be outdated"

Evidence

Rather than restate the README's claims, every row marked implemented was checked against source in this tree and cited as file:line. Two examples:

// repo/repo.v:1796 — private-repo read gate, enforced at 43 call sites
fn (app &App) can_read_repo(ctx Context, repo Repo) bool {
	if repo.is_public {
		return true
	}
	return ctx.logged_in && app.user_has_repo_read_access(ctx.user.id, repo)
}

// issue.v:54 — multiple assignees, contradicting the assumption of a single one
struct IssueAssignee {
	issue_id   int @[unique: 'issue_assignee']
	user_id    int @[unique: 'issue_assignee']
}

Two findings are worse than plain gaps, because a settings screen implies the feature is one click away:

Wiki has a setting and no feature.

// repo/repo.v:49
disable_wiki            bool

// repo/repo.v:69
fn (r &Repo) wiki_enabled() bool {
	return !r.disable_wiki
}

update_repo_features persists the toggle, but a repository-wide search for wiki finds no wiki content model, no wiki route, and no wiki template. An owner can switch it on and find nothing.

Topics is a counter. repo/repo.v:23 declares nr_topics int @[skip], repo/repo_template.v:87 formats it, but no route assigns topics, so the explore page shows a number that is always zero:

/topics : 0    (no route matches)

The document separates what is genuinely present, a work queue ordered by dependency rather than a wish list, what is deliberately out of scope so it is not re-raised as a defect, and a final Not verified section. That last section exists deliberately: rows I could not check are listed rather than guessed, so an unexamined row is never mistaken for a negative one. It currently holds multiple PR reviewers, the topic UI surface, the batch-issue UI, and git hook extensibility.

Merge Danger

Door: Two-way. Documentation only; no product code, schema, or data. Reverting removes a file and one README sentence.

Blast Radius: None at runtime. The risk is reputational rather than technical: this document states Gitly does not have features the README currently implies, including the two stubs above. That is accurate and was verified, but it is a claim about the project's completeness, so it is the kind of change a maintainer may want to soften rather than have land cold.

The list of missing features is ordered by dependency, which means the first four items (wiki, topics, in-repo code search, comment reactions) are the cheap ones. It does not estimate effort, and it does not commit to any of it — it is a work queue, not a roadmap.

[medvednikov]

Thanks for separating verified, missing, and not-verified capabilities. A few entries need tightening before this is relied on as a factual parity matrix: (1) Low RAM / CPU usage is marked implemented/verified solely on V, single binary, which is not evidence for actual resource consumption; mark it unverified or add measurements. (2) Multiple reviewers is ? not verified inside Implemented and verified, and ? is absent from the legend; move it to Not verified. (3) Telemetry says Gitly has absolutely none, whereas the repository implements telemetry ingestion and storage (platform_operations_routes.v/platform_operations.v). If the intended claim is no outbound usage tracking/phone-home, say that specifically and distinguish it from the product telemetry feature. These changes would make the matrix more trustworthy without changing the scope.

[metif12]

Done - addressed in the latest push.