www-plugins/adobe-flash is currently the only package depending *unconditionally* on corefonts (or a particular alternative) instead of virtual/ttf-fonts. The cause of this appearst to be bug 227217, but I tried the video in that bug with liberation-fonts + corefonts instelled and with only libertation-fonts installed, and couldn't see any difference. (However, the timer was always displayed on the bottom, so maybe actually adobe-flash itself has changed since then). If you have an example that virtual/ttf-fonts is really not sufficient, I suggest to make at least the dependency optional, e.g. with USE=truetype as all current ebuilds of wine do. Frankly, I (and probably many other users) do not want to install a font just to get perhaps some timer displayed in some exotic videos.
It seems to me that adobe-flash should depend on media-fonts/liberation-fonts or media-fonts/corefonts. Without liberation-fonts flash did not work correctly. With liberation-fonts it worked, at least after having them enabled via eselect fontconfig. I did not try without enabling. Uninstalling freefonts-ttf (current dependency) didn't have a noticable effect. I didn't install corefonts, but see bugs #302176 and #227217. My personal opinion is that adobe-flash is crippled without either liberation-fonts or corefonts. It is needed, for instance, for playing farmville on facebook, or for the annoying iTunes banner on youtube. Thanks, Henry
(In reply to comment #0) > www-plugins/adobe-flash is currently the only package depending > *unconditionally* on corefonts (or a particular alternative) > instead of virtual/ttf-fonts. According to this page, that's not exactly true: http://tinderbox.dev.gentoo.org/misc/rindex/media-fonts/corefonts but I agree that it is not very common, and in many cases it really is undesirable to depend on corefonts directly. > The cause of this appears to be bug 227217, but I tried the video in that > bug with liberation-fonts + corefonts instelled and with only libertation-fonts > installed, and couldn't see any difference. (However, the timer was always > displayed on the bottom, so maybe actually adobe-flash itself has changed > since then). To see a difference you would have to uninstall both corefonts *and* liberation-fonts, as each provide a nearly glyph-identical set of fonts. Without either fontset installed, you would start to notice misaligned, truncated, or missing strings in your flash applets due to flash designers assuming that when they ask for "Arial" they get a font that is sized the same as Microsoft's arial.ttf. > If you have an example that virtual/ttf-fonts is really not sufficient, Bug #302176 and as you mentioned, bug #227217. I think both bugs demonstrate fairly concretely that the proper dependency for adobe-flash is: || ( media-fonts/liberation-fonts media-fonts/corefonts ) And not: virtual/ttf-fonts > I suggest to make at least the dependency optional, e.g. with USE=truetype > as all current ebuilds of wine do. USE=truetype should mean: "Adds support for FreeType and/or FreeType2 fonts" since this is the description of the global USE flag, and I don't think that's would fit what you are suggesting here. I suspect I would have to come up with some local USE flag that would mean "don't install any extra fonts"... Maybe I could use "minimal" for that, but to eliminate a single font that seems a bit overkill. > Frankly, I (and probably many other users) do not want to install a font just > to get perhaps some timer displayed in some exotic videos. Liberation-fonts is only 1.58M in size (28 files), so I don't think that's a huge burden to carry. Compare it to, say, libflashplayer.so itself which is 9.2M
(In reply to comment #2) > (In reply to comment #0) > > www-plugins/adobe-flash is currently the only package depending [...] > > According to this page, that's not exactly true: You are right. Now I forgot how I checked it; probably, I used "equery d" (which checks only my installed ~1200 packages) and made a mistake when grep'ing the remainder. > To see a difference you would have to uninstall both corefonts *and* > liberation-fonts, as each provide a nearly glyph-identical set of fonts. If the ebuild dependency were (at the time of opening the bug) || ( liberation-fonts corefonts ) I would have tried that. (Actually, in this case, I would not have opened this bug, since as a liberation-fonts user, I would not have realized the depdency on corefonts). > due to flash designers assuming that when they ask for "Arial" > they get a font that is sized the same as Microsoft's arial.ttf. Such problems will always exist: There are also many web pages which will never work without java (or even never without IE). However, this is no reason to make java a mandatory dependency for all webbrowsers (or even dump all webbrowsers under linux), although some might consider this as crippled. > USE=truetype should mean: "Adds support for FreeType and/or FreeType2 fonts" > since this is the description of the global USE flag, and I don't think > that's would fit what you are suggesting here. It is in analogy with wine's (and probably some other ebuild's, too) usage of that flag. I am not so familiar with gentoo's policy, but my impression was that it would be preferred to have a global useflag which has perhaps only a rather related meaning - the specific meaning can always be explained in metadata.xml. However, "minimal" is also good, IMHO. > Liberation-fonts is only 1.58M in size Actually, I am already happy with liberation-fonts as alternative dependency. So from my egoistic point of view, since the alternative dependency is now liberation-fonts, this bug can be closed. :)
(In reply to comment #3) > Actually, I am already happy with liberation-fonts as alternative dependency. > So from my egoistic point of view, since the alternative dependency is now > liberation-fonts, this bug can be closed. :) I'm glad we agree :) If I get many more complaints about the too-specific nature of the font requirement, I will consider adding a "minimal" use flag to skip them. I know I fixed the font dependency bug just after you filed this one, so I hoped that was the main reason for the confusion. Thanks very much for the report.
