some code-paths may leak it. ok stsp@

it's still used a bit later, so defer the free'ing at the next loop iteration or at function end.

in case of a failure between when it's allocated and when it's released it might get leaked otherwise. ok and semplification stsp@

ok stsp@

ok stsp@

getchange() in diff3.c takes any line that starts with an integer and treats it as a hunk header. It doesn't actually require ed script instructions, and traditional ed script diff output would break it. Thus, changing Got's diff3 to rely on plain diff output mode paves the way for implementing traditional ed-script style output in diff.git. Patch by Tom Jones

in the loop of match_loose_object we allocate a string per directory entry that in some case it was free(3)'d before `continue' or `goto', but not always. Instead, use a more common idiom. ok stsp@

This removes the gcc ternary if extension (includes follow-up diff.git fix for missing '\' in wrapped macro 7a38b3a33f8b124c9c01da154dfdf711d328b01d)

ok stsp@

if strdup fails we fail to free the querystring Found by clang' scan-build.

dead store found by clang' scan-build.

In 95a4a5a1e8d0952ee0d2523dc31df7170d80e0b3 I forgot to fix the links for the favicons and the web manifest.

problem introduced in commit 625e5896fc9ecf87ccfc92ad2a65cd3be58f73c0

ok op@

got_get_repo_tags needs some tweaking to how tag_commit is allocated too. ok stsp@

all very unlikely. part of a bigger diff that's ok stsp@

drop: - a tautological s[strlen(s)] == '\0' - a few duplicates checks - a lone if (error) when it's always NULL - an extra NULL check part of a bigger diff that's ok stsp@

fcgi_parse_params parses fastcgi parameters into a list. (This is a leftover from slowcgi where that list is later used to populate the environment of the CGI process.) However, this list is never looked at and its memory never released, so just drop it. Make the matching on fastcgi parameters name strictier by checking also that the length is the one we expect; otherwise we might pick up parameters with the same prefix string (i.e. FOO vs FOO_WITH_SUFFIX) While here turn some bcopy into memcpy and simplify some if-nesting too. Fix the reading from an un-initialized pointer that I introduced in a previous commit. ok stsp@

- don't special case ".got": if it's a work-tree gotwebd will just fail soon enough. Suggested by stsp@ - avoid an strdup. - drop `opened' to make it easier to follow what's going on. ok stsp@

ok op@

ok tracey@