Stashes are currently stored using the reflog in a given repository. This is an interesting and novel way to handle them, but there is no way to easily move a set of stashes across machines. For example, groups of stashes cannot be bundled, pushed, or fetched. Let's solve this problem by allowing users to import and export stashes to a chain of commits. The commits used in a stash export contain two parents: one which is the pointer to the next exported stash (or to an empty commit with no parents if there are no more) and the second is the stash commit that would normally be stored in the reflog. Original thread at message-ID: <20220310173236.4165310-1-sandals@xxxxxxxxxxxxxxxxxxxx> Changes from v6: * Add Phillip's sign-off to the last patch. * Use `commit_list` for tracking commits. * Use reflog entry walker. * Fix some commit messages for improved legibility. * Rephrase some error messages for precision. * Drop the patch that exposes `read_complete_reflog` since it is no longer necessary. Changes from v5: * Rename `parse_revision`. * Remove extra call to `free_stash_info`. * Fix parsing of existing commit. * Add more validation of imported stash commits. * Add more tests for improved validation of imported stash commits. * Explicitly cast `items.nr` and make the iteration counter an `ssize_t` to avoid casting problems. * Don't require a trailing `\n\n` in commit messages. * Use `read_complete_reflog` to walk reflogs. * Be more defensive when using `lookup_commit_reference`. * Apply parts of Phillip's patches for improved robustness. * Update commit message to explain additional use cases. * Use `OPT_STRING` for `--to-ref`. Changes from v4: * Fix another use of oid_array. * Fix various memory leaks. * Fix a segfault which appeared after a rebase. * Use strstr for commits since we don't need to worry about NUL. * Added some additional tests. * Verify the ident values we're using to avoid using bad values. * Various other code cleanups. * Rebase on `master`. Changes from v3: * Fix strbuf handling to avoid leaks and generally be more sensible. * Make use of the error return code more often. * Use oid_array. * Tidy various parts of the code and fix long lines. * Simplify tests using git tag. * Shorten and tidy tests. * Add an additional test covering the base commit OID and importing and exporting empty stashes. Changes from v2: * Fix uninitialized strbuf. * Avoid C99-style initializations. Changes from v1: * Change storage format as suggested by Junio. * Rename to GIT_OID_GENTLY. * Remove unnecessary initializations. * Use ALLOC_GROW_BY. * Ensure completely reproducible exports. * Avoid size_t. * Various other code cleanups. brian m. carlson (4): object-name: make get_oid quietly return an error builtin/stash: factor out revision parsing into a function builtin/stash: provide a way to export stashes to a ref builtin/stash: provide a way to import stashes from a ref Documentation/git-stash.adoc | 29 ++- builtin/stash.c | 460 ++++++++++++++++++++++++++++++++++- hash.h | 1 + object-name.c | 6 +- t/t3903-stash.sh | 94 +++++++ 5 files changed, 577 insertions(+), 13 deletions(-)
