In addition to the previous commit [c450903bc2] that plugged fetch and send leaks, free what would be the path entry's path member, which is leaked when attempting to add a duplicate. And in the fetch case, free refname and id when got_pathlist_insert() returns err. ok stsp@

Patch by op@, who is at the wrong computer right now

Patch by: Kyle Ackerman

Patch by Bradley Taunt ok op@

Check for duplicate items added to pathlists, and free associated data in case we were attempting to add a duplicate. Patch by: Kyle Ackerman

Calling fmemopen(3) with a size of zero leads to an "Invalid argument" error. One symptom of this issue are failing gotd notifications: gotd[48539]: render notification: fmemopen: Invalid argument Add an email notification test case which triggers the above error. Reported by Mischa Peters

tested and ok stsp@

ok op@

done mechanically with $ find . -iname \*.[ch] -exec sed -i -E 's/[[:space:]]+$//' {} + with some revert for bundled files and a manual tweak for got-fetch-http.c

As reported by Johannes Thyssen Tishman, 'got stage -p $binaryfile' iterates each hunk, and produces unexpected behaviour when answering 'y'. Instead, present the standard "Binary files differ" message and only allow staging (or unstaging or reverting) the entire binary file. Includes tests for stage -p but unstage and revert -p tests are still needed. tweak + ok stsp@

Fixes spurious "branch on server has a different ancestry" errors from 'got send' when the repository history contains merge commits.

For the first 13 iterations of the loop we read behind the start of the array. Instead of just adding a range check, rewrite the code to split the loop in two so it's easier to follow. Reported by Timo Myyrä ok stsp@

The first `nmeta - max_base_candidates` were not freed in `got_pack_create`. Patch by Kyle Ackerman ok op@

Some parts of the code rely on staged_blob_id being NULL in certain cases. Restore this behaviour while keeping the segfault bug fixed.