* wip
* basics are working
* turret case is handled
* better fighter beam handling
* missing commit?
* handle missile spawn
* clean up, commenting and ssb handling
* random in sphere and per burst rot counts for "share random" flag
* clean up, clarity and comments
* accidental whitespace changes
* bugfix
* NULL -> nullptr
* more efficiency/clarity
* review tweaks and bugfixes
Here's a bug that's been around ever since version 3.6. If the subsystem names were not essentially the same, the function wouldn't do the expected string comparison.
The script-eval SEXP evaluates all arguments as individual script commands, which is not ideal because most FREDders expect the arguments to be concatenated. Since the script-eval-block SEXP handles its arguments this way, it makes sense to hide the script-eval SEXP in FRED to prevent confusion.
Commit 6c2ebab01c (#3439) added a `max_line_length` argument to `split_str()`, but one call had the extra argument added in the wrong spot, and so a previously-pointless-but-harmless `ignored_char` of 0 became a `max_line_length` of 0, resulting (ultimately) in an exception in the unicode code when the iterator tried to decrement at the beginning of a ship's description. This gets rid of the argument entirely, rather than reordering them, because ignoring null characters doesn't really do anything anyway.
I've also added `Assertion`s to check `max_line_length` that should prevent this specific problem from occurring again in the future.
* Validate user-provided values passed to Random::next()
* return low value on invalid range
* call LuaError() instead of Warning() in rand32() script function
* Added a byte limit for line length in split function
* Fix call error
* Ádd missing limits import
* Change hardcoded values into defines
* Cleanup default argument
If the `for-players` collection has no valid arguments, don't short-circuit, just to be on the safe side.
Don't short-circuit at all for the `for-ship-*` sexps, because ships can arrive and depart.
#3271 changed the logic of `stuff_token_list()` in such a way that the supplied index being used to access an array was no longer skipped if it was invalid; naturally, this means that this warning was now being inevitably followed up by an out-of-bounds array access. Adding a `!skip_this_entry` check should hopefully fix the problem while still allowing the loadout count parsing code below it to run, which was presumably the reason the `continue` statement was replaced with the `skip_this_entry` variable in the first place. Fixes#3405.
To be more user-friendly, and to accommodate buggy retail missions, make the one-argument version of `is-ship-visible` default to the first player in multiplayer mode.. Also, change the warning to an mprintf.
There were a few assertions in the SEXP code that would trip if an invalid variable name were specified in a SEXP, but nothing that actually stopped a Release build from performing an out-of-bounds array access if someone made a typo in notepad. Since, as far as I could tell, everything should go through `get_sexp_text_for_variable()` first, I changed its `Assert()` into an `Error()` that should hopefully ensure that we avoid ever doing anything with `Sexp_variables[-1]`.
This moves the ## to # escape sequence logic from `mission_process_alt_types` to its own function where it can be used elsewhere (such as for ship class and weapon class display names that need to do the same thing). Also, this introduces a !! escape sequence for exclamation points that is needed for exclamation points to be used in scripts, as @naomimyselfandi pointed out.
When various SEXP functions related to ammo were consolidated back in June (in commit 54409ec928), certain safety catches meant to avoid having too much ammo or the wrong ammo capacity after changing weapons were lost. This commit adds code that should be functionally equivalent.
Consolidate all the functions which parse lists of data. They're almost the same except for the data itself, so use templates.
As for the templated function which handles the list, boost the robustness by adding additional checking and error messages.
While skimming the differencs in jg18's branch of PR #3253...
https://github.com/jg18/fs2open.github.com/tree/feature/sexp-containers-revised
...I noticed a TODO that is almost certainly a genuine bugfix. As such, it deserves its own PR. The line in question was last modified in 2012 by karajorma, and the bugfix should test for the special-arg in the entire actions tree rather than just the first action.
JohnAFernandez