| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
| |
This is probably not the usual cause of such problems in WebTiles,
though:
* This doesn't usually happen on servers except when
manually-triggered rebuilds result in the same branch being built
twice in a row AND a PNG has been resized.
* This problem won't go away on reloading the page, since the
coordinates won't be up-to-date in *any* file.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
This also adds redone backgrounds for equipped (and cursed equipped)
items to match the recently added new item slot background.
|
|
|
|
|
|
|
|
| |
This allows the image and the code to be built independently; thus their
dependencies can be (and now are) checked separately, preventing code
from being rebuilt when only tile images have been updated (as opposed
to the dc-*.txt definitions of the images, which are a dependency for
both the image ant the code).
|
|
|
|
| |
This fixes incomplete cleans, etc.
|
| |
|
|
|
|
|
| |
If you changed, for example, just dc-item.txt, it resulted in a misbuild
since headers were out of sync (dc-main.txt %includes dc-item.txt).
|
| |
|
| |
|
|
|
|
|
|
|
| |
In this case, it will produce enums but no images, avoiding a dependency on
SDL. Since production of images is very deeply intertwined with rest of
processing, this change is a nasty hack all around, having it process 0x0
dummy images all over. Doesn't seem to explode, though.
|
| |
|
| |
|
|
|
|
|
| |
It seems like this was the only Makefile in the whole tree that didn't
check for changes in CFLAGS between runs.
|
| |
|
|
|
|
|
|
|
|
| |
greatzebu's scripts and his change in cc3e1d6d relied on host and target
being the same, which is true only for native builds. When passing CXX
from the outside you need to specify which one you mean.
After this commit, you can override them separately.
|
| |
|
| |
|
|
|
|
|
| |
With per-arch contrib builds, this could be implemented by building SDL,
SDL-image and libpng twice, but that'd need work.
|
|
|
|
|
|
|
|
| |
This fixes compilation errors when compiling for a different architecture
without a manual "clean-contrib", will make it possible to fix failures due
to libpng and sdl-image (which need to be available both for host (tile
data) and for target (the game itself)), and may allow reducing unnecessary
recompilations if we cease to pass flags to contribs.
|
|
|
|
|
|
| |
The Mac tiles build can trip over MacPorts SDL and libpng being incompatible
versions and compiled for the wrong architecture, so allow the user to force
use of contrib SDL and libpng.
|
|
|
|
| |
This should also allow us to draw clouds beneath monsters.
|
|
|
|
| |
Signed-off-by: Steven Noonan <steven@uplinklabs.net>
|
|
|
|
|
|
| |
Due to the set of coloured floors and walls, it was getting perilously
large. The tiledef-dngn.h file now just indirectly calls the correct
feat/floor/wall functions depending on which tile index is used.
|
|
|
|
|
|
|
|
|
|
| |
On Mac OS X, pkg-config gave a very odd result for --libs, which included
both the static _and_ shared SDL libs. This resulted in numerous multiply
defined symbols. The program warned that it would use either the locally
defined symbol or the shared library one, but which one was at best a
random choice. This could cause very unpredictable runtime behaviour.
Signed-off-by: Steven Noonan <steven@uplinklabs.net>
|
|
|
|
|
|
|
|
|
|
|
| |
Some 'which' implementations print errors to stdout instead of
stderr, which can give false positives. Fix this by redirecting
all output of 'which' to /dev/null, and checking the return value
instead.
Thanks to JoachimSchipper on Mantis for this suggestion (Issue #926).
Signed-off-by: Steven Noonan <steven@uplinklabs.net>
|
|
|
|
|
|
|
| |
Tested with a local Windows msysgit build and CDO's mingw cross-compile. (Thanks to Napkin!)
Signed-off-by: Janne Lahdenpera <felirx@gmail.com>
Signed-off-by: Eino Keskitalo <evktalo@users.sourceforge.net>
|
|
|
|
| |
Just another ".d file in a subdirectory" problem.
|
| |
|
|
|
|
| |
Signed-off-by: Steven Noonan <steven@uplinklabs.net>
|
|
|
|
| |
Signed-off-by: Steven Noonan <steven@uplinklabs.net>
|
|
|
|
| |
Signed-off-by: Steven Noonan <steven@uplinklabs.net>
|
|
|
|
|
|
| |
They were rebuilding over and over. Now they're not.
Signed-off-by: Steven Noonan <steven@uplinklabs.net>
|
|
|
|
| |
Signed-off-by: Steven Noonan <steven@uplinklabs.net>
|
|
Signed-off-by: Steven Noonan <steven@uplinklabs.net>
|