Summary
README.md tells a new user to clone from smartwatermelon/dotfiles, which is
the pre-migration location. It works today only because GitHub redirects; a
fresh clone lands with an origin remote pointing at a name this repo no longer
uses, which then produces confusing gh failures (see below).
README.md:159:git clone git@github.com:smartwatermelon/dotfiles.git ~/Developer/dotfiles
README.md:164:git remote add origin git@github.com:smartwatermelon/dotfiles.git
Both should be git@github.com:twistedmelonman/dotfiles.git.
Why this is worth fixing even though the redirect works
A checkout whose origin says smartwatermelon/dotfiles pushes fine, but
gh pr create fails with a misleading error:
GraphQL: Head sha can't be blank, Base sha can't be blank, Head repository
can't be blank, No commits between twistedmelonman:main and
smartwatermelon:claude/<branch>, Head ref must be a branch, not all refs are
readable (createPullRequest)
gh resolves the base from the redirect target but the head from the literal
remote name, treats them as two different repos, and reports it as "no commits
between" rather than as a stale-remote problem. Hit while opening #325; worked
around with an explicit --head twistedmelonman:<branch>. Anyone following the
README gets the same failure with no obvious cause.
Fixed locally in this working copy with git remote set-url, but a fresh clone
by the documented instructions reproduces it.
Scope: deliberately narrow
This repo has ~49 files mentioning smartwatermelon. Most of them must NOT
change:
smartwatermelon is still a live org owning ~25 repos
(github-workflows, dev-env, homebrew-tap, pr-review, scripts, ...).
Verified via the API: smartwatermelon/github-workflows and
smartwatermelon/dev-env resolve to themselves, not to a redirect. Only
~33 repos moved to twistedmelonman. References to repos that stayed are
correct as written.
bash/gh-wrapper.sh lines 275 and 520 list
smartwatermelon | nightowlstudiollc | twistedmelonman as the in-org
allowlist. That is correct and load-bearing — three names, one person. Do
not touch.
- Test fixtures use the old name deliberately;
test-gh-wrapper-gh-token-precedence.sh specifically asserts that a stale
smartwatermelon GH_TOKEN fails closed. Rewriting those would delete the
regression coverage.
- Issue citations (
smartwatermelon/dotfiles#135 and similar, ~30 files)
are cosmetic. GitHub redirects them. Churning 30 files of comments to
renumber references buys nothing and would bloat the blame history.
So this issue covers the two README.md clone/remote lines only.
Optionally also pre-commit/README.md:25, which links to
smartwatermelon/dotfiles/issues/190 — same cosmetic category as the rest of
the issue citations, listed here only because it sits in a README a human
reads rather than in a code comment.
Note
.git/config in this working copy has no uchg flag set (stat -f %Sf
returns -), so the historical immutability problem that made remote edits
fail appears to be resolved. No production code in the repo sets uchg; the
only chflags reference is a cleanup trap in
bash/tests/test-git-env-isolation.sh.
Summary
README.mdtells a new user to clone fromsmartwatermelon/dotfiles, which isthe pre-migration location. It works today only because GitHub redirects; a
fresh clone lands with an origin remote pointing at a name this repo no longer
uses, which then produces confusing
ghfailures (see below).Both should be
git@github.com:twistedmelonman/dotfiles.git.Why this is worth fixing even though the redirect works
A checkout whose origin says
smartwatermelon/dotfilespushes fine, butgh pr createfails with a misleading error:ghresolves the base from the redirect target but the head from the literalremote name, treats them as two different repos, and reports it as "no commits
between" rather than as a stale-remote problem. Hit while opening #325; worked
around with an explicit
--head twistedmelonman:<branch>. Anyone following theREADME gets the same failure with no obvious cause.
Fixed locally in this working copy with
git remote set-url, but a fresh cloneby the documented instructions reproduces it.
Scope: deliberately narrow
This repo has ~49 files mentioning
smartwatermelon. Most of them must NOTchange:
smartwatermelonis still a live org owning ~25 repos(
github-workflows,dev-env,homebrew-tap,pr-review,scripts, ...).Verified via the API:
smartwatermelon/github-workflowsandsmartwatermelon/dev-envresolve to themselves, not to a redirect. Only~33 repos moved to
twistedmelonman. References to repos that stayed arecorrect as written.
bash/gh-wrapper.shlines 275 and 520 listsmartwatermelon | nightowlstudiollc | twistedmelonmanas the in-orgallowlist. That is correct and load-bearing — three names, one person. Do
not touch.
test-gh-wrapper-gh-token-precedence.shspecifically asserts that a stalesmartwatermelonGH_TOKEN fails closed. Rewriting those would delete theregression coverage.
smartwatermelon/dotfiles#135and similar, ~30 files)are cosmetic. GitHub redirects them. Churning 30 files of comments to
renumber references buys nothing and would bloat the blame history.
So this issue covers the two
README.mdclone/remote lines only.Optionally also
pre-commit/README.md:25, which links tosmartwatermelon/dotfiles/issues/190— same cosmetic category as the rest ofthe issue citations, listed here only because it sits in a README a human
reads rather than in a code comment.
Note
.git/configin this working copy has nouchgflag set (stat -f %Sfreturns
-), so the historical immutability problem that made remote editsfail appears to be resolved. No production code in the repo sets
uchg; theonly
chflagsreference is a cleanup trap inbash/tests/test-git-env-isolation.sh.