docs/reference/commands/verify.btft

[table class="topnav"]
[tr]
[td class="logotab"]mp[e]
[td][see name="index"]index[e][e]
[td][see name="guide/index"]guide[e][e]
[td][see name="reference/index"]reference[e][e]
[e]
[e]
 
[h1]mp verify[e]
 
[h2]synopsis[e]
 
[code]
mp verify [pkg[:slot] ...]
[e]
 
[h2]description[e]
 
a sanity sweep, not an install-time check: with no arguments, checks every port in every configured repo; given one or more names, checks just those. for each, confirms the port directory exists, its [code]pkg.conf[e] parses, and it has a resolvable install phase (a new-style [code]install:[e] phase or class, or a legacy [code]add.sh[e]) AND a resolvable remove phase ([code]remove:[e]/[code]del.sh[e]). prints [code]ok: name.[e] or [code]err: name, reason.[e] per port, and exits nonzero if anything failed.
 
this does NOT resolve dependency graphs, check for cycles across the whole tree, or catch a real build failure -- it's "does this port even parse and have both phases", the cheapest possible tree-wide check. [see name="reference/commands/dryrun"]mp dryrun --force all[e] is the heavier sibling that actually resolves everything.
 
[h2]examples[e]
 
[code]
mp verify
ok: cat/subcat/figlet.
err: cat/subcat/broken, bad pkg.conf.
[e]
 
[h2]see also[e]
 
[see name="reference/commands/dryrun"]dryrun[e], [see name="reference/testing"]reference/testing[e] (the regression suite this same invariant gets exercised by).