On Wed, 9 May 2018 09:33:07 -0700 Marc MERLIN <marc_...@merlins.org> said:
> I'll change the title for the different discussion. > > First a question: does the version number mean what I think it means, or > does it really not mean much? > My guess/expectation is that > 0.21.0 start with new features > 0.21.x gets tweaks and bug fixes > 0.21.11 is the last from that featureset branch and the most stable of that > branch (at least ideally) yes. the micro version (last version number changing above) means basically bug fixes only (or relevant changes that are equivalent to a bug fix). > 0.22.0 start over with brand new features and bugs > etc.. 0.22 should have all the fixes 0.21.xxx has as the fixes in 0.21.xxx will have been cherry-picked/backported from 0.22. 0.22 may introduce new bugs we don't know about, but unless people try code in git before it becomes 0.22 and report issues, we don't know. bugs that we see get fixed. bugs we don't, do not. bugs that are reported may get fixed if we can reproduce them or possibly divine what it might be (but if its not reproducible by developers then it's impossible to fix without help from the affected user - by sending them patches or having them try attempted fixes to verify is needed). > Does that sound correct? > > > > 1) never had time to even get anything building until today > > > 2) 0.22 has too many different bugs for me to want to keep trying to use > > > it. > > > > well then we are at an impasse. nothing is going to change or be fixed > > because there is no way to know what to change or why, so you will be on > > 0.21 forever and we'll move on. sorry. we've been through this before. :( > > We have. The problem here is that I've literally never, not once, > upgraded E because I wanted new features. I've only ever upgraded it > because I absolutely had to for a reason or another (bugs, > incompatibilities). > I'd have stayed with 0.18.5 forever if it weren't for it not working > with chrome anymore, and so forth. > Hell, I'd likely still be on 0.16 if I could ;) > However, every time I've upgraded E, I got unwelcome to very unwelcome > new bugs, making me regret my upgrade :-/ then try git master before releases - use a different ~/.e directory (or back up ~/.e - tar it up and when going back to the previous stable wipe ~/.e and untar it ... though do this without e running). yes - this requires a build env that you have to set up once. it's installing some -dev pkgs. it's not hard. i've done it many times. i've done it on horribly slow machines too and it's not that time consuming. maybe 10-20mins on a slow box (re-running configure to get to the next error each time. no need to re-do autogen.sh as long as configure was generated ... for efl. for e it's meson now so same thing and its lightning fast vs configure). if you don't participate, then this happens. e has so many options that it is impossible to test them all in all combinations as well as all possible use cases with all possible apps and all possible machines with differing speeds and thus race conditions and all the possible applications and their own behaviour patterns etc. etc. the only way to do that is to have many users participate. it's the only way to scale such QA. we have pages documenting even what packages to install to compile (they aren't quite right but they get you 90% of the way there). how to compile etc. has docs on enlightenment.org. it isn't a laborious time sink unless you choose to make it one and do it the hard way and not use these docs. :) if you stick to an old e and you have issues you will be left behind. that's reality. we can't support old versions. even supporting one is enough work and even that doesn't get all the bug fixes back-ported. really only some that are clear and simple and easily cherry-picked, and perhaps the odd one that needs work. > Yes, you're going to hate me for saying that, sorry :) > Basically E has done what I needed for a long time. It's not adding any > features I need (although I'll admit that the pulse sound slider per > window is actually a nice new feature in 0.22). > > I know I've come off as complaining and not contributing. Sadly, it's > because my WM has to work, it's not a hobby for me, it's like X, it has > to work because all my other work depends on it. my wm has to work too - all day. i do development in it... and mine does. solidly. it's not crashing or doing weird focus things like for you. i use git master as it rolls all day every day. the thought that "i must use something old and stable otherwise it will never work" is wrong. one problem with a release is developers move away from that release to git master instantly and if they find bugs they fix them in git master and then backport (maybe - see above). but people using the latest release will not be using what developers use. that is normal and to improve this we should do more frequent releases, but that wouldn't work for you as you would never upgrade unless forced, so what can we do for you but never do any development or changes to e ever again except bug fixes only, and even then you wouldn't update unless pushed. :( > I actually spend a lot of time contributing patches to projects, > finding bugs, reporting them in details if I can't fix them myself, and > help test the new code. I also maintain my own projects and code. But to > do all this, I need linux, Xorg, and E to work. Neither I'm interested > in beta testing because I don't have another machine I use just to test > them, I only use them on my production machines where they have to just > work. git master just works for me. all the time. but as above. you can have it installed and compilable and test it from time to time and if you find issues, report them when you find them. you can check out a specific revision if you find one revision with a bug and report what revision it was that caused it and go back 1 step and wait for a fix (which should be pretty fast). > Last week, I show up for a work meeting, I open my laptop, everything is > hosed/hung. I try to restart E, it won't restart. I forget why but that > time I actually had to restart X. Time lost, people looking at me. > I restart X and chrome and as chrome starts, one tab start blazing noise > and I can't quickly find which one. The E speaker volume slider is > broken on that start, I can't change the volume or mute quickly. > I excuse myself, leave the room, go debug all this outside while looking > like an idiot. i just hit the mute key... :) my kbd has one. i've never had the mixer slider not actually work on me (unless actually busy working on that code messing with it as i've done sometimes). > Lesser but not uncommon scenario: I open my laptop in the plane with all > my stuff pre-loaded. If E wedges itself and I can't fix it/restart it > without crashing X (usually I succeed thankfully), I lose all my > preloaded work and then no wifi to recover during the flight, I'm hosed. this just doesn't happen to me. i don't know what kind of things you run and do on your machines but i do not see this. but if e is edged you should figure out why and report - e.g. a backtrace. but what on earth? you lose work? how? what kind of workflow do you have that forces your work to live in very volatile process state and not be saved to disk? that's just crazy to work that way. it's like doing your work along on a tightrope over a pit of flames... it's asking for trouble. i have seen e "stop rendering". it's not wedged but alive just not rendering (and i have never caught WHY), but ctrl+alt+end and it's back to normal as it's not hung. i have also seen on intel drivers SOMETIMES e hang inside a gl driver call - e is stuck before the opengl driver hangs and stops moving. this is solved with a "killall -SEGV enlightenment" to force it to crash and hit f1 to restart... but this is not something we can fix as it's to do with mesa (if it's intel which is where i saw this), or kernel drm/kms drivers. i don't know which. so i can fix it without crashing x... i unblock e by forcing a crash of e. but i need a backtrace to know why/where it's hung to know what to do. > So, that's why linux, X, and E just have to work and why I don't to ever > upgrade E unless I must, as it's pretty much been pain and > disappointment every time (except that usually I was also leaving > another version with other pain and other disappointment). > I so so so wish for an E branch that just stabilizes and works forever, > or almost forever. I however know you don't have a release team like > linux with a person responsible for maintaining the 4.9.x kernel for X > years. > But having people follow the bleeding edge and git forever is not a > solution for people like me either. I hope you understand. we're an an impasse. if you have issues and don't help figure out what is wrong they will not be fixed. if you don't want to be part of the solution... you're part of the problem. :( it's a little harsh, but i;'m trying to encourage you to actively help. > In my ideal life, I want neither new features nor new bugs, just fewer > bugs until they're few enough that I can live with them. Until I'm > forced to switch to wayland, or some other chrome comes up with new > stuff that breaks older versions of E and then I'm hosed again. reality is this is coming. and if you don't want to participate then you will get to deal with the problems. > > 0.22 has this fixed... so either move to 0.22 and try what i suggested or go > > but it has new features and new bugs, and I need/want neither :) and there will be 0.23, 0.24, 0.25 and so on and you'll be eventually left so far behind the only response will be one word: "update" :). > > we don't try to put stuff in e or efl that is broken for us - in our > > testing, but not all distros and hardware are the same. they contain > > different versions > (...) > > Yeeh, I totally understand that. I totally understand that the problems > I see may not be E's fault, but they appear after I upgrade E. At work they may be e's problem too, but without some debugging we don't know. i have seen issues that are both e or efl's fault or something else. many times. > when stuff breaks, you revert the last thing changed, regardless of > whose fault it really is in the end. Same with E. but the last thing you changed isn't a single change. it's a vast amount of changes. going from 0.22 to 0.21 or 0.21 to 0.20 is a haystack of changes and you are looking for a needle. which is why my suggestion is to have a stable release that works and also participate with git master to help weed out problems before a new stable release so you can move to that and sit on that happily for a while and every now and again move forward to the git master (tar up ~/.e as above) and see if all is ok etc. ... participate. reality is that you are our qa department. like it or not. you choose not to participate but then you will see issues that maybe only you see and not have any way for them to be fixed. > To be fair, I tried to upgrade Xorg + intel drivers and all I got for > that effort is that now xscreensaver is unable to turn my screen dark > before locking. Sigh :( (not E's fault of course) > > If I don't get new features I don't need, I won't get new bugs I don't > need either ;) you get to keep all the current bugs then... :) > Makes sense? > > Any chance we can have a stable E branch? It's super vexing if I'm we have a stable e branch per release. see out git repo. every release gets a stable branch and commits are backported to it. we only backport to the last release stable branch because the delta to master is the least and patches apply the most easily. that is why i have to push you to the last stable release because that *IS* the stable e branch you ask for. it's already done. > forced to go to 0.22 to get fixes to know bugs in 0.21, and 0.22 is > totally not stable for me, nor is it apparently going to be anytime soon > (and even less so without my helping you, which isn't very realistic due > to what I explained above). > This is why I built 0.21.11 with wishful thinking that it would be the > last set of bugfixes to a known stable branch. Seems that it's not > though :-/ > > Marc > -- > "A mouse is a device used to point at the xterm you want to type in" - A.S.R. > Microsoft is to operating systems .... > .... what McDonalds is to gourmet > cooking Home page: http://marc.merlins.org/ | PGP > 7F55D5F27AAF9D08 > -- ------------- Codito, ergo sum - "I code, therefore I am" -------------- Carsten Haitzler - ras...@rasterman.com ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ enlightenment-users mailing list enlightenment-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-users
