ok millert@

ok millert@

follow the "proper secure idiom" described in the CAVEATS section of printf(3). reminded by tb@ and millert@

"looks good to me" millert@

Cm sign matches the markup used in the ssh-keygen(1) man page. Patch by Josiah Frentsos

patch by Josiah Frentsos

don't free the querystring, it's already done later by fcgi_celanup_request. ok tracey@

ok jamsek

As per stsp's TODO item, make tog respond to key presses while "loading..." history. Initial testing in src.git appears to work as intended. ok stsp@

If a child blame view has been opened, halfdelay() is reset to a 1 second refresh rate, so the index in the header lags behind subtree navigation as we set the index based on the selected tree entry, which isn't updated till that refresh cycle completes. Don't wait for the selected_entry to update; instead, compute the index from the first displayed entry and the currently selected line index. improved by and ok stsp@

Previously, we only handled 'G' in log_move_cursor_down() when all commits had not yet been loaded; move the case where all commits are already loaded into this routine too, and simplify by handling both cases the same. This change was prompted by Mikhail's observation of a redundant assignment in the previous 'G' handling in log_move_cursor_down(). ok stsp@

When at the end of the log view such that the currently selected line is greater than nlines/2, add--don't assign--the page down value to the selected index so we don't jump up when the user expects tog to scroll down. While here, rework log page down logic for a smoother UX in horizontal split mode by moving the selection cursor down rather than scrolling when 'd' or 'f' is entered if the last commit is already on screen. ok stsp@

When checking for specific sys/queue.h declarations, ensure to add the header file (sys/queue.h) to search, otherwise nothing will be found.

ok kn, tracey

it was broken in 7ecc454 because now we're emitting tags with multiple `class' attributes (by mistake) and only the first one is used. However, instead of fixing the HTML `class' attribute, use a CSS selector to alternate the background colors. ok tracey@

This adds some buffering to fcgi_gen_binary_response so that we don't end up sending too many (very) small fastcgi records. While here also make fcgi_send_response non-void and make its errors "sticky". ok tracey@

HTML IDs are meant to be unique across the page. While here fix two small errors in the generated HTML (one missing closing div and one extra.) ok tracey@

Needed for -portability reasons. OK op@

Someof the socket code has been factored out of gotwebd and put centrally so that platform-specific targets can be created, without the need of heavily modifying the central socket code to contain numerous ifdefs.

I don't think this assignment to sin_len is needed at all, because sin_len is already being set via got_sockaddr_inet[6]_init() during the configuration stage. Tested successfully on both fcgi socket types (unix / tcp).

ok stsp@

got_error_path uses get_custom_err to get a statically allocated error allocated on a buffer that's used a ring. This is nice and simple, but the drawback is that thees custom errors shouldn't be used "for too long" because they might get overwritten later. cmd_info is one offender: change it to store a "simple" error and later re-use it to construct the per-path error. OK and tweaks stsp@