saumini/count
A lightweight Laravel package for efficient Eloquent counting. Get accurate totals for models and relationships without loading full collections, making it useful for dashboards, list cards, and search results where performance and simple code matter.
Architecture fit. The critical finding for a TPM is that the marketing copy and the code describe different products. The description sells a general-purpose counting library for "controllers, services, and views"; the repository is nsaumini/nova-field-count and the artifact is a Laravel Nova presentation field — a Field subclass plus a Vue component registered through Nova::script(), rendering an integer sourced from an Eloquent withCount / {relation}_count attribute. There is no schema, migration, config, queue, job, or write path. The addressable surface is therefore one paid, closed-source admin panel, not the platform. If the requirement lives anywhere other than a Nova index or detail view, this package contributes nothing and the answer is withCount in the application layer. The binding constraint is the installed Nova major — not Laravel, not PHP.
Integration feasibility. High and conditional, gated on two cheap verifications rather than a debate. First, data source: the field must read a {relation}_count attribute already loaded via withCount on the Resource's indexQuery; if instead it resolves per row, it is an N+1 dressed as a convenience and the value proposition inverts. The README is the entirety of the documentation, so this is answered by reading resolve/resolveUsing in src/, not by trusting prose. Second, assets: Nova field packages typically ship a pre-built dist/ served through Nova's asset route, meaning no npm/Vite chain is introduced into the host app — a material de-risking factor for a team without front-end tooling, but one to confirm on a staging deploy rather than assume.
Technical risk. Moderate, dominated by lifecycle rather than defect risk, and therefore a procurement problem more than an engineering one.
nsaumini/nova-field-count versus package saumini/count, plus a description that misdescribes scope, is precisely the shape of a typo-squat or dependency-confusion mistake. Verify the exact Packagist name and publisher against the repository before any composer require.{relation}_count attribute renders 0. Operators will read that as real data — "zero open tickets" on an account with twelve — which erodes trust in the admin panel faster than a visible crash ever would. For a TPM this is the risk that matters: a rendering bug costs a page refresh, a wrong number costs the credibility of every number on the screen.Key questions. All are answerable in under half a day of source reading, and the answers retire the remaining speculation:
^4.0 only, or ^5.0 too?withCount on the index query, or issue a query per row?Stack fit. Laravel 9–11, PHP 8.1+, Nova 4.x (verify 5.x). Strictly complementary to the existing admin surface — nothing changes in routing, auth, tenancy, storage, or queues. That absence of infrastructure coupling is the strongest argument for running a cheap pilot instead of holding a design review: the cost of being wrong is an afternoon, and the cost of deciding slowly is the same afternoon spent.
Migration path. Purely additive and fully reversible. composer require, one line in a Resource's fields(), withCount added to that Resource's indexQuery if the field requires it, deploy. Rollback is composer remove plus deleting one line — no data migration, no user-visible disruption, no runbook. Frame this explicitly to stakeholders: the decision is not a platform bet, it is a reversible experiment, and the irreversibility that usually justifies a committee simply isn't present.
Compatibility. Pin the version and treat Nova as the binding constraint. Run composer why-not laravel/nova before any Nova upgrade and record this package as a known upgrade dependency on the Nova upgrade checklist. The roadmap consequence is the part executives need to hear: the package carries a small recurring verification cost on every Nova upgrade, not a one-time adoption cost — and because breakage surfaces in the browser rather than in CI, it will not be caught by the test suite.
Sequencing.
NovaField, or Text::make()->displayUsing() over a withCount attribute.The central finding for roadmap purposes is that the build-vs-buy calculus here is nearly symmetric — the package's value is convenience and a slightly nicer UI, not capability. There is no defensible reason to accept lifecycle risk on a component you could own outright in an afternoon, unless the hours saved are genuinely redeployed this sprint. The decision to defend to stakeholders is honestly "vendor ~100 lines versus write ~100 lines," and either answer can be correct.
Maintenance. Near-zero runtime maintenance; the real cost is periodic re-verification against Nova majors. Budget roughly half a day per major upgrade to confirm the field still registers and renders, and make it a blocking verification step on the Nova upgrade checklist — not an afterthought discovered by an operator. Vendoring (copy the class and Vue component into app/Nova/Fields/) converts an external lifecycle dependency into owned code and drops that verification to "does our own field still compile."
Support. There is no vendor to escalate to: one maintainer, no issue-response history, MIT license with no SLA. Every support path terminates internally. That is acceptable only because the blast radius is internal — a rendering regression costs an operator a refresh, not a customer — and because the fallback is an afternoon of work. Any requirement that imposes an SLA or a security-response obligation on this dependency should be treated as disqualifying rather than negotiated.
Scaling. The scale axis is not data volume in the abstract but query shape per rendered row: 500 rows × a correlated subquery against a multi-million-row related table is a visibly slow page on a screen operators use all day. Mitigations are architectural, not field-level — summary tables, materialized counters, or a read model — and they sit outside this package's scope. Where counts must be sortable or filterable at scale, this field is the wrong instrument and the honest answer is a metrics workstream. The staging smoke test with query logging is the whole of the scaling diligence and should not be skipped.
Failure modes. Ranked by how much damage each does to trust in the admin panel:
0 as though it were fact. Highest severity, lowest detectability, and invisible to CI.The mitigation is the same for all four and is cheap: a single staging smoke test with query logging, plus one assertion in the pilot review that the rendered number matches a hand-run count() query.
Ramp-up. Effectively zero for engineers already fluent in Nova fields — the mental model is "wrap withCount in a Field," and the code surface is ~100 lines. The ramp-up cost that actually matters is organizational, not technical: building the habit of adding this package to the Nova upgrade checklist, and of asking "is this a count, or is it a metric the business needs to trust?" If the requirement drifts toward sums, averages, distinct or conditional aggregates, trends, time series, or cross-record comparison, that is the signal to escalate to a real metrics workstream — metric definition, ownership, and reconciliation — rather than accrete features onto a display field. Flag it and re-scope; the field is the trivial last step of that project, not the project itself.
How can I help you explore Laravel packages today?