Is it possible add "nacl" use flag into chromium ebuild?
Why?
(In reply to comment #1) > Why? To exclude building huge independent nacl toolchain
NaCl is not that popular yet, but when web pages start relying on it, having it optional may break them. I don't think the toolchain is really so bad. You should only need to build it once. Compared to cumulative compile time of frequently updated chromium, toolchain compilation time should be a minority. By the way, the upstream's disable_nacl gyp switch may break from time to time.
*** Bug 406211 has been marked as a duplicate of this bug. ***
(In reply to comment #3) > NaCl is not that popular yet, but when web pages start relying on it, having > it optional may break them. Is Gentoo going to force Flash and Java also then? > By the way, the upstream's disable_nacl gyp switch may break from time to > time. They support it. Why can't Gentoo allow it on versions where it works?
(In reply to comment #5) > Is Gentoo going to force Flash and Java also then? No, but this is different: NaCl is part of the package in question. > > By the way, the upstream's disable_nacl gyp switch may break from time to > > time. > > They support it. Why can't Gentoo allow it on versions where it works? Who told you it's supported? I've seen at least 3 various breakages, and was talking in person to people responsible for it (and also I'm an upstream developer). Those various switches are not really supported, or in other words are supported until they break.
(In reply to comment #6) > (In reply to comment #5) > > Is Gentoo going to force Flash and Java also then? > > No, but this is different: NaCl is part of the package in question. How so? It might not use a standard plugin API, but it's still an optional external package. > > > By the way, the upstream's disable_nacl gyp switch may break from time to > > > time. > > > > They support it. Why can't Gentoo allow it on versions where it works? > > Who told you it's supported? I've seen at least 3 various breakages, and was > talking in person to people responsible for it (and also I'm an upstream > developer). Those various switches are not really supported, or in other > words are supported until they break. thakis and traxmac; I'm mainly taking about Chromium releases, which are going to have a const value of "is it broken" at least.
Created attachment 304441 [details, diff] Make Native Client optional, controlled by "nacl" USE flag (enabled per default) Here is a patch against the current portage tree to make NaCL optional. I have tested all the release versions, and all of them work fine without NaCL.
(In reply to comment #8) > Created attachment 304441 [details, diff] [details, diff] > Make Native Client optional, controlled by "nacl" USE flag (enabled per > default) > > Here is a patch against the current portage tree to make NaCL optional. I > have tested all the release versions, and all of them work fine without NaCL. No, we're not going to accept this patch. Sorry. I really hope I have explained the rationale, and at some point someone is going to say "no" to Yet Another USE Flag.
USE flags is half the point of Gentoo. In this case, you're forcing non-free software on people unnecessarily, so it's also a violation of Gentoo's Social Contract.
Phajdan gets to pick what patches he accepts; I'm certainly not in a position to judge. If Chromium was the only browser available in Gentoo I might be swayed by your contract argument; but there are many available. -A
(In reply to comment #11) > If Chromium was the only browser available in Gentoo I might be > swayed by your contract argument; but there are many available. Chromium is the only *stable* browser... :/
To make it worse, NaCL isn't even an option for non-x86 (ARM). So this is a case where the feature is forced on x86, and not available on other arches. Totally awkward.
I second the notion of having nacl as a use flag. This is exactly what useflags are for (in my book). If the ebuilds don't make use of the very fabric of gentoo, we can all go use debian or something. If a potential "nacl" useflag is enabled by default, the dont-care-user wont even notice and the do-care-user can decide for himself to do something you don't agree with (namely exclude nacl from his chromium). "I don't think it's so bad" is really not an argument to forbid disabling it imho. Your opinion is valuable, but that doesn't mean it is the good to force it on other people. "the upstream's disable_nacl gyp switch may break from time to time." - this is indeed a good point, but only people who deliberately disabled nacl would see this; there are a number of ways for an ebuild to communicate to the user that a configure-breakage is (or might be) due to disabling nacl. "NaCl is part of the package in question" - with this argument, you'd have to remove 50% of all useflags, since they just enable/disable features that are part of the "package in question". "at some point someone is going to say "no" to Yet Another USE Flag" - as I tried to state above, useflags are an essential part of what makes gentoo great! I love with being able to tweak my system to what I really want. mplayer has about 90 useflags, chromium has 5 - don't worry it wont get overcrowded so soon... As I see the matter, your arguments are mostly invalid, whereas Luke-Jr has excellent points. Please, heed my call, do not force things on your users, let them choose, because choice that's what gentoo is all about in my book! PS: It is a good strategy if you want your users to learn write their own ebuilds, I give you that, but that's not what an ebuild maintainer should be after if you ask me...
*** Bug 429474 has been marked as a duplicate of this bug. ***
FWIW, I've been maintaining a NaCL-optional branch of the ebuilds in my personal overlay (layman -a luke-jr). I don't update it too often, though. Bug 429474 is interesting; is NaCL still a problem on ARM? Older chromium ebuilds forced it disabled on ARM as a special case, but that was removed in the newer ones...
Created attachment 343566 [details, diff] Updated patch to make NaCL support optional Can we get this reopened yet? I've been maintaining my overlay with this one patch updated in chromium ebuilds for over a full year now, but it's getting tiresome! Over the past year, I have never seen a stable build where it doesn't work either - it seems the whole "upstream's disable_nacl gyp switch may break from time to time" claim is either history or FUD. People obviously want the choice, and it does no harm to those who want nacl support.
Created attachment 403796 [details] ebuild for building chromium-43.0.2357.73 with NaCl Why don't you let the users make their choice? Just set the USE flag not enabled by default!
(In reply to Kelvin Ng from comment #18) > Created attachment 403796 [details] > ebuild for building chromium-43.0.2357.73 with NaCl > > Why don't you let the users make their choice? Just set the USE flag not > enabled by default! This bug was about optionally *disabling* NaCl back in the days when we still had it enabled unconditionally. If you have managed to make NaCl work with recent versions of chromium, please attach a patch to bug 541610. I don't want to resurrect this old, largely irrelevant bug report.
