thank you for your feedback!
i'll classify my responses by solution status:
will talk with upstream
-----------------------
On Tue, Jul 30, 2013 at 04:51:28PM +0200, Paul Wise wrote:
> About the fonts stuff; The best would be if the program used SDL Pango
> instead or as an alternative to SDL TTF. Other font rendering systems
> that automatically discover system fonts and cope with multi-script
> strings by falling back on other fonts are also OK. If upstream isn't
> OK with switching to SDL Pango, please send them a patch to use
> fontconfig to detect where the font files are located, falling back on
> runtime or compile-time configured fonts. An example of how to do this
> stuff is in chromium-bsu, which uses different rendering libs but the
> principles are the same. The GLC backend is an example of the SDL
> Pango style and the FTGL backend is an example of the SDL TTF style.
given that the game only displays its own static strings and usually
ships the font file, i'd assume upstream will stay with sdl ttf, but i
asked anyway.
> Could you ask upstream; What is the origin of the numbers in
> polygons.cpp? Were they created by hand? Were they converted from some
> other format and hard-coded into the file? Are they maintained in some
> other format? Are they not maintained at all?
my impression was that it's a mixture of copy/pasted and one-shot import
from somewhere else, possibly an ad-hoc process -- but i've asked for
clarification here too.
i don't think it's a problem
----------------------------
> Why is -O3 in CXXFLAGS? I would suggest sticking to the Debian default
> which is set by dpkg-buildflags (-O2) and not setting CXXFLAGS in the
> Makefile.
that's the default value upstream ships in the make file. as it is
packaged, i patched the make file to only -O3 if the flags are not given
from outside, so it shouldn't affect debian. (the original makefile
didn't respect any environment variables.)
> Please add --parallel to the dh arguments.
the build process is a single gcc invocation. --parallel wouldn't do
anything, would it?
> P: hyperrogue: no-upstream-changelog
given how rare upstream changelogs are outside of top-of-class
upstreams, i sometimes wonder why this is even a check.
ok / done
---------
> Please use imagemagick or similar at build time to create PNG and XPM
> icons (for the Freedesktop and Debian menus) from the upstream MS
> Windows icon. The XPM should be 32x32 and the PNG one should be the
> maximum size available in the MS Windows icon.
done. should the .ico file remain in /usr/share/pixmaps? (i assume no.)
> Please forward the remaining patches, manual page and desktop file
> upstream if you haven't already.
the patches were already forwarded, man page and desktop file are now.
> You may want to run wrap-and-sort -sa.
i wasn't aware of that tool, good to know, thanks and done.
> The copyright years in debian/copyright are slightly wrong.
oups. (only looked at the file all the other files referred to). fixed.
> CXXFLAGS missing (-fPIE)
i enabled hardening, blhc doesn't complain any more.
> desktop-file-validate:
>
> debian/hyperrogue.desktop: error: value
> "hyperbolic;escher;noneuclidean" for locale string list key "Keywords"
> in group "Desktop Entry" does not have a semicolon (';') as trailing
> character
i blame the poor resolution to #277441.
just kidding, now i know of the tool, fixed and done.
> bfbtester:
>
> Lots and lots and lots of core dumps due to SIGABRT.
on my box, i got segfault crashes for single argument tests.
i'm not sure exactly why, but they are related to hyper segfaulting when
no DISPLAY variable is set (failure to check for SDL_Init return code).
now runs flawlessly; probably the x server was under too heavy load, sdl
didn't get proper windows back, and behaved as if the x server was gone.
i'm under the impression that this kind of thing might happen again --
i've already spotted another zero pointer dereference
(03-worm-segfault.patch), which might be indicative of a general
sloppyness with respect to zero pointers and exit codes, but my stance
on this is that it's sufficient quality for a game which neither has
elevated privileges nor reads untrusted data.
---------
i hope to get the package in release-able shape when i receive zeno's
responses to my inqueries concerning the typesetting system and further
sources; i'll report back then.
chrysn
--
To use raw power is to make yourself infinitely vulnerable to greater powers.
-- Bene Gesserit axiom
Attachment:
signature.asc
Description: Digital signature