Commit Briefs
simplify lib/send.c reference handling
Reorganize the reference validation and pathlist generation by removing the reflist and building a pathlist directly. The pathlist entries record the object id in their extra data pointer, which also allows several redundant reference lookups to be skipped. This will eventually simplify sending target reference names that do not match the local repo by adding another parameter to insert_sendable_ref for a remote reference. This remote name will be added to the pathlist, but validation and object id lookups will continue to be performed with the local reference. ok jamsek
significantly reduce the amount of code linked into gitwrapper
By moving got_serve_parse_command() from lib/serve.c into lib/dial.c as got_dial_parse_command(), we can significantly reduce the amount of symbols gitwrapper depends on indirectly. As a downside, gotsh now needs to link to dial.c. But it only uses the same parsing routine, and any other routines in dial.c would likely cause pledge violations in gotsh if used. No functional change.
rename lib/sha1.c to lib/hash.c
It will soon grow functions to deal with sha256 too. stsp@ agrees.
include sha2.h too where sha1.h is included
In preparation for wide sha256 support; stsp@ agrees. Change done mechanically with find . -iname \*.[cy] -exec sam {} + X ,x/<sha1\.h>/i/\n#include <sha2.h>
unbreak send.sh regress from recent ref-delta changes
fix argument order typo in flags passed to got_pack_create()
gotadmin pack: add a -D flag to force using ref-delta
Intended mostly for the regress suite, so we'll be able to test also the ref-delta code paths. ok stsp@
got: minor refactor of got_pathlist_free() API
Accept flag parameter to optionally specify which pointers to free. This saves callers looping through the list to free pointers. ok + fix stsp@
got send: show server error
Print the error message reported by the remote server when failing to update a branch (for e.g. because of a server-side check.) Reported by gonzalo@, with help and ok stsp@.
let callers of got_pack_create() configure rate-limiting of progress reporting
Needed by future gotd(8), where progress reports will be sent to a network socket, rather than a local terminal.
move got_opentempfd() call out of got_pack_create()
Future gotd(8) needs to run got_pack_create() in a chroot environment, so we can no longer open new temporary files inside got_pack_create(). ok op@
make got_pack_create() write to a file descriptor instead of a stdio FILE
The old code required a seekable output file. This conflicts with requirements of future gotd(8), which will write pack file data to network sockets. ok op@
Correctly track if we sent more data since the last time we looked.
With the narrow scope of bytes_sent_cur it would always be zero, and we would always update progress. Found by llvm's scan-build (Dead store). OK stsp
fix regression where 'got send -T' failed if same tag already exists on server
Problem reported and fix tested by Omar Polo.
move duplicated dial_ssh() and dial_git() functions into a common file
These functions are used by 'got send' and 'got fetch' in order to open network connections to a server. Move them into new file lib/dial.c and declare relevant functions in got_dial.h and lib/got_lib_dial.h. No functional change.
make realloc_ids() malloc-like and do not overallocate
Let realloc_ids() take as argument the number of entries to allocate. Do not allocate an extra chunk. ok stsp
drop realloc_ids() initialization calls, one of which had a typo
realloc_ids() is called again in the loops that assign to the ID arrays, before the first assignment. ok stsp
do not send a pack file when 'got send' is only deleting branches
The git protocol spec says the client MUST NOT send a pack file if the only command used is 'delete'. Fixes 'got send -d' against Github's server which closed the session upon receiving the empty pack file we sent. This problem wasn't caught by regression tests since git-daemon does accept an empty pack file. Problem reported by jrick.