Commits


Xr gotweb(8) from got(1)


typo


long overdue update of AUTHORS lists in man pages; based on commit history


regress test SSH key revocations ok stsp@


move 'got init' command to 'gotadmin init' This functionality is better suited for gotadmin because it is technically a server-side repository operation when we consider the 'got' tool as the client-side tool. I have plans to move 'got import' into gotadmin as well. ok op@


add missing revoked_signers grammar to got.conf(5) ok stsp@


fix potential type mismatch between format specifier and argument Cast printf argument of type time_t to long long to match the %lld format specifier on platforms where this might not be the case.


bump version number


CHANGES for 0.73


document TOG_VIEW_SPLIT_MODE


include system headers required for got_date.h directly, not indirectly


sync distfile list


document SSH-based signing and verification in got(1) and got.conf(5)


in tests, verify that tag -V behaves like tag -l plus signature verification


got tag -V is like tag -l with verification, fix option-conflicts accordingly


remove parts of tag_create_ssh_signed which just duplicate the tag_create test


add -v option to 'got tag' usage string


add option-conflict handling for 'got tag' -s and -V options


add -V option to 'got tag' usage string, and replace underscore with a hyphen


check for specific chars instead of using isspace(3) Reminded by naddy and stsp; it was missing a cast to unsigned char to prevent issues on archs with signed chars and was too broad anyway. While here, drop an extra check immediately after. ok stsp@


got_date.h: add explicit #includes Don't rely on implicit/"hidden" includes, but rather add them to the header file so their inclusion doesn't result in missing values. Although not explicitly a functional change, this does help -portable. This ensures size_t/time_t are always present. OK @jrick


remove duplicate test_parseargs call spotted by op@


fix echo/printf order and actually run the test


Use pipe() which is a more understood syscall than pipe2() which doesn't exist on MacOS, for instance. Since we we're passing in 0 to pipe2(), this mean no fcntl() flags were being sent. As such, it's the same syscall as pipe() which also has the added benefit that it's more portable. committing on behalf of thomas with my ok


whitespace fix