Re: [gentoo-user] Re: ncurses: reductio ad absurdum
On Sat, 29 Aug 2015 17:17:29 -0400, Fernando Rodriguez wrote: $ sudo emerge -lp portage These are the packages that would be merged, in order: Calculating dependencies... done! [ebuild R] sys-apps/portage-2.2.20.1 What am I doing wrong? I guess because it's a rebuild and it only shows relevant changes? Yes, it shows changelog entries since the installed version. I have a cron script that syncs, does emerge -f @world and then emails me the output from emerge -luDp @world. -- Neil Bothwick Sure, we just route the main sensor through Data's cat. pgpaMJt29TGlK.pgp Description: OpenPGP digital signature
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
On Saturday, August 29, 2015 6:13:30 PM Dale wrote: Fernando Rodriguez wrote: On Saturday, August 29, 2015 10:48:16 AM Dale wrote: Todd Goodman wrote: * Philip Webb purs...@ca.inter.net [150828 18:35]: 150828 Rich Freeman wrote: To really appreciate git you should understand git objects and their references, what a commit, tree, and blob are. Also, the whole copy-on-write concept and content-hashing concept. I used to think git looked really complicated until I sat through a 1 hr talk that focused mostly on the data model. Once you understand the data model, you understand everything. That doesn't take a lot of time. It does take a moderate amount of time learning the right things. They're not found in the manpages. Like I said, beautiful design, horrible interface. So is there a Gentoo doc -- Wiki, presumably -- explaining to users -- users, not dev's or Git addicts -- the essentials of Git, so that they can readily update using it ? If so, I'm willing to see if I can use it ; if not, I would suggest it sb a top priority for dev's to write. You don't *need* to know anything about git to update using it. Just change your /etc/portage/repos.conf/gentoo.conf as Rich outlined (and move away your rsync'd /usr/portage or wherever your portage tree goes.) Then when you emerge --sync (or emaint -A sync, etc.) it will sync via git and emerge will work as always. Now if you want to do more or just want to learn more about git then that's different. Todd I think what we are talking about is viewing things like the changelogs and such, which are currently not synced with the tree. Or did we change to some other topic and I missed it? I tracked back to Alan Mackenzie's split of this thread . Dale :-) :-) It's probably easier to do this: # cd /usr/portage # rm -r * # git clone repo-uri . Then do the repos.conf changes. That way you don't have to worry about portage doing a shallow clone. If you already did it then just unshallow it as Rich pointed. Then to view the logs just: #cd /usr/portage/cat/pkg #git log . Then 'git show first few digits of commit hash' to view a commit diff. You can use git use dev-vcs/tig if you find it easier though I thought it was pretty useless so it only lasted about 10 secs. in my system. So basicly the only change is that instead of: # less ChangeLog (or whatever you use to read logs) You'll do: # git log . Actually, I use eix-sync to sync my tree. However I do it, I want it done within the usual setup and commands. Given the bumps we've already seen, I'm not wanting to change that just yet. Let the devs work out some of the kinks first. I use eix-sync too, it just calls emerge --sync so it's the same. I'm not in a hurry to switch the main tree to git either. If they bring change logs to rsync I'll stick with it as long as it's supported. Git will just be more wasteful of disk space and has other potential problems that rsync doesn't. I think it's great of version control but not so much for this. Oh, I use Kwrite to read the changelogs. If I'm stuck in a console, nano, head or cat works. Well, it did in the past anyway. May not now tho. Dale :-) :-) -- Fernando Rodriguez
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
On Saturday 29 Aug 2015 04:32:48 Dale wrote: walt wrote: On Fri, 28 Aug 2015 15:29:20 -0400 Rich Freeman ri...@gentoo.org wrote: I used to think git looked really complicated until I sat through a one hour talk that focused mostly on the data model. Once you understand the data model, you understand everything. That doesn't take a lot of time. Does that talk happen to be available on youtube or some equivalent? I'm a git fanboi in spite of its failings. I've been using git since Linus and Larry McVoy divorced (amicably, claims Linus) over where the kernel source repo would reside. That amicable divorce spawned the development of git in the first place (says Linus). I'm no expert on git, but 'git bisect' has allowed me to file countless (for countless = than the number of cc's of vino in my glass) bug reports over the years. I see that all the gentooers who replied to my post have been lingering in this mailing list for years. (You are Old Farts, by my definition.) All gentoo Old Farts are here because we are gentoo addicts and not one of us could abandon gentoo even if we wanted to. Including me. That may help me too. It took me a while to have a sorta understanding of how Gentoo is set up ebuild wise and such and now I feel like I did back in 2003. Having it on youtube or something would be really good. I did find this tho: https://www.youtube.com/watch?v=73I5dRucCds I'll try to watch it later. Don't. You'll waste an inordinate amount of time listening to the drone of someone's uncontrollable verbiage, which contains less that 1% meaningful information. I suspect that 10 minutes should be enough to explain what git is, what github is, how you use them. Unfortunately, any videos I found over a 10 minute youtube search didn't provide me with anything useful to share. :-( -- Regards, Mick signature.asc Description: This is a digitally signed message part.
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
On Saturday, August 29, 2015 10:48:16 AM Dale wrote: Todd Goodman wrote: * Philip Webb purs...@ca.inter.net [150828 18:35]: 150828 Rich Freeman wrote: To really appreciate git you should understand git objects and their references, what a commit, tree, and blob are. Also, the whole copy-on-write concept and content-hashing concept. I used to think git looked really complicated until I sat through a 1 hr talk that focused mostly on the data model. Once you understand the data model, you understand everything. That doesn't take a lot of time. It does take a moderate amount of time learning the right things. They're not found in the manpages. Like I said, beautiful design, horrible interface. So is there a Gentoo doc -- Wiki, presumably -- explaining to users -- users, not dev's or Git addicts -- the essentials of Git, so that they can readily update using it ? If so, I'm willing to see if I can use it ; if not, I would suggest it sb a top priority for dev's to write. You don't *need* to know anything about git to update using it. Just change your /etc/portage/repos.conf/gentoo.conf as Rich outlined (and move away your rsync'd /usr/portage or wherever your portage tree goes.) Then when you emerge --sync (or emaint -A sync, etc.) it will sync via git and emerge will work as always. Now if you want to do more or just want to learn more about git then that's different. Todd I think what we are talking about is viewing things like the changelogs and such, which are currently not synced with the tree. Or did we change to some other topic and I missed it? I tracked back to Alan Mackenzie's split of this thread . Dale :-) :-) It's probably easier to do this: # cd /usr/portage # rm -r * # git clone repo-uri . Then do the repos.conf changes. That way you don't have to worry about portage doing a shallow clone. If you already did it then just unshallow it as Rich pointed. Then to view the logs just: #cd /usr/portage/cat/pkg #git log . Then 'git show first few digits of commit hash' to view a commit diff. You can use git use dev-vcs/tig if you find it easier though I thought it was pretty useless so it only lasted about 10 secs. in my system. So basicly the only change is that instead of: # less ChangeLog (or whatever you use to read logs) You'll do: # git log . -- Fernando Rodriguez
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
Todd Goodman wrote: * Philip Webb purs...@ca.inter.net [150828 18:35]: 150828 Rich Freeman wrote: To really appreciate git you should understand git objects and their references, what a commit, tree, and blob are. Also, the whole copy-on-write concept and content-hashing concept. I used to think git looked really complicated until I sat through a 1 hr talk that focused mostly on the data model. Once you understand the data model, you understand everything. That doesn't take a lot of time. It does take a moderate amount of time learning the right things. They're not found in the manpages. Like I said, beautiful design, horrible interface. So is there a Gentoo doc -- Wiki, presumably -- explaining to users -- users, not dev's or Git addicts -- the essentials of Git, so that they can readily update using it ? If so, I'm willing to see if I can use it ; if not, I would suggest it sb a top priority for dev's to write. You don't *need* to know anything about git to update using it. Just change your /etc/portage/repos.conf/gentoo.conf as Rich outlined (and move away your rsync'd /usr/portage or wherever your portage tree goes.) Then when you emerge --sync (or emaint -A sync, etc.) it will sync via git and emerge will work as always. Now if you want to do more or just want to learn more about git then that's different. Todd I think what we are talking about is viewing things like the changelogs and such, which are currently not synced with the tree. Or did we change to some other topic and I missed it? I tracked back to Alan Mackenzie's split of this thread . Dale :-) :-)
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
* Philip Webb purs...@ca.inter.net [150828 18:35]: 150828 Rich Freeman wrote: To really appreciate git you should understand git objects and their references, what a commit, tree, and blob are. Also, the whole copy-on-write concept and content-hashing concept. I used to think git looked really complicated until I sat through a 1 hr talk that focused mostly on the data model. Once you understand the data model, you understand everything. That doesn't take a lot of time. It does take a moderate amount of time learning the right things. They're not found in the manpages. Like I said, beautiful design, horrible interface. So is there a Gentoo doc -- Wiki, presumably -- explaining to users -- users, not dev's or Git addicts -- the essentials of Git, so that they can readily update using it ? If so, I'm willing to see if I can use it ; if not, I would suggest it sb a top priority for dev's to write. You don't *need* to know anything about git to update using it. Just change your /etc/portage/repos.conf/gentoo.conf as Rich outlined (and move away your rsync'd /usr/portage or wherever your portage tree goes.) Then when you emerge --sync (or emaint -A sync, etc.) it will sync via git and emerge will work as always. Now if you want to do more or just want to learn more about git then that's different. Todd
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
* Dale rdalek1...@gmail.com [150829 11:49]: Todd Goodman wrote: * Philip Webb purs...@ca.inter.net [150828 18:35]: 150828 Rich Freeman wrote: To really appreciate git you should understand git objects and their references, what a commit, tree, and blob are. Also, the whole copy-on-write concept and content-hashing concept. I used to think git looked really complicated until I sat through a 1 hr talk that focused mostly on the data model. Once you understand the data model, you understand everything. That doesn't take a lot of time. It does take a moderate amount of time learning the right things. They're not found in the manpages. Like I said, beautiful design, horrible interface. So is there a Gentoo doc -- Wiki, presumably -- explaining to users -- users, not dev's or Git addicts -- the essentials of Git, so that they can readily update using it ? If so, I'm willing to see if I can use it ; if not, I would suggest it sb a top priority for dev's to write. You don't *need* to know anything about git to update using it. Just change your /etc/portage/repos.conf/gentoo.conf as Rich outlined (and move away your rsync'd /usr/portage or wherever your portage tree goes.) Then when you emerge --sync (or emaint -A sync, etc.) it will sync via git and emerge will work as always. Now if you want to do more or just want to learn more about git then that's different. Todd I think what we are talking about is viewing things like the changelogs and such, which are currently not synced with the tree. Or did we change to some other topic and I missed it? I tracked back to Alan Mackenzie's split of this thread . Dale That's fine. I was addressing Philip wanting to update via git. If you want more than just synching and emerge then learn the one or two git commands you need. As Rich said, just use git log and git diff if you want to see changes. It's not hard, just different. Todd
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
On Saturday, August 29, 2015 9:53:10 PM Neil Bothwick wrote: On Sat, 29 Aug 2015 12:35:35 -0400, Fernando Rodriguez wrote: So basicly the only change is that instead of: # less ChangeLog (or whatever you use to read logs) You'll do: # git log . Or if emerge is updated to use git too, instead of emerge -l you'll use emerge -l ;-) I didn't knew that one :) But it doesn't seem to work even with rsync: $ sudo emerge -lp portage These are the packages that would be merged, in order: Calculating dependencies... done! [ebuild R] sys-apps/portage-2.2.20.1 What am I doing wrong? -- Fernando Rodriguez
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
On Saturday, August 29, 2015 5:10:15 PM Fernando Rodriguez wrote: On Saturday, August 29, 2015 9:53:10 PM Neil Bothwick wrote: On Sat, 29 Aug 2015 12:35:35 -0400, Fernando Rodriguez wrote: So basicly the only change is that instead of: # less ChangeLog (or whatever you use to read logs) You'll do: # git log . Or if emerge is updated to use git too, instead of emerge -l you'll use emerge -l ;-) I didn't knew that one :) But it doesn't seem to work even with rsync: $ sudo emerge -lp portage These are the packages that would be merged, in order: Calculating dependencies... done! [ebuild R] sys-apps/portage-2.2.20.1 What am I doing wrong? I guess because it's a rebuild and it only shows relevant changes? -- Fernando Rodriguez
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
On Sat, 29 Aug 2015 12:35:35 -0400, Fernando Rodriguez wrote: So basicly the only change is that instead of: # less ChangeLog (or whatever you use to read logs) You'll do: # git log . Or if emerge is updated to use git too, instead of emerge -l you'll use emerge -l ;-) -- Neil Bothwick I thought I saw the light at the end of the tunnel... but it was just some sod with a torch bringing me more work! pgp6HrnqmhXhF.pgp Description: OpenPGP digital signature
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
On Sat, Aug 29, 2015 at 12:35 PM, Fernando Rodriguez frodriguez.develo...@outlook.com wrote: It's probably easier to do this: # cd /usr/portage # rm -r * # git clone repo-uri . The only issue with this is that all the files end up being owned by root. I'd just create /usr/portage, chown portage:portage /usr/portage, and then let emerge --sync take care of the rest. To cut down on replies, I'm not sure what the emerge --sync behavior is if you have changes in the tree. I suspect that as long as they don't conflict they'll still sync, but in general you shouldn't leave uncommitted changes lying around /usr/portage. It is just convenient to be able to tweak packages, get them cleaned up, and then generate patches. -- Rich
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
On Saturday, August 29, 2015 2:19:23 PM Rich Freeman wrote: On Sat, Aug 29, 2015 at 12:35 PM, Fernando Rodriguez frodriguez.develo...@outlook.com wrote: It's probably easier to do this: # cd /usr/portage # rm -r * # git clone repo-uri . The only issue with this is that all the files end up being owned by root. I'd just create /usr/portage, chown portage:portage /usr/portage, and then let emerge --sync take care of the rest. That is true, I didn't think of that. However, emerge --sync must run as root because that's how I've done it for my local overlay and I just noticed that even my /usr/portage (I'm still using rsync) is owned by root:root with the exception of distfiles which is owned by root:portage. Probably because at one point it got corrupted after syncing and I copied it from a network machine. Fixing it now. To cut down on replies, I'm not sure what the emerge --sync behavior is if you have changes in the tree. I suspect that as long as they don't conflict they'll still sync, but in general you shouldn't leave uncommitted changes lying around /usr/portage. It is just convenient to be able to tweak packages, get them cleaned up, and then generate patches. At least for overlays portage does a git pull. It may pull the changes from the server but it will not even try to merge them (it can't since there's no commit to merge) so it will tell you to either git checkout the file to discard the changes or commit them. Even if you do commit them I believe it will ask you for a merge commit message so it's not something that portage will do automatically. Best to create a testing branch for your changes and checkout the master branch before syncing again. -- Fernando Rodriguez
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
Todd Goodman wrote: * Dale rdalek1...@gmail.com [150829 11:49]: Todd Goodman wrote: * Philip Webb purs...@ca.inter.net [150828 18:35]: 150828 Rich Freeman wrote: To really appreciate git you should understand git objects and their references, what a commit, tree, and blob are. Also, the whole copy-on-write concept and content-hashing concept. I used to think git looked really complicated until I sat through a 1 hr talk that focused mostly on the data model. Once you understand the data model, you understand everything. That doesn't take a lot of time. It does take a moderate amount of time learning the right things. They're not found in the manpages. Like I said, beautiful design, horrible interface. So is there a Gentoo doc -- Wiki, presumably -- explaining to users -- users, not dev's or Git addicts -- the essentials of Git, so that they can readily update using it ? If so, I'm willing to see if I can use it ; if not, I would suggest it sb a top priority for dev's to write. You don't *need* to know anything about git to update using it. Just change your /etc/portage/repos.conf/gentoo.conf as Rich outlined (and move away your rsync'd /usr/portage or wherever your portage tree goes.) Then when you emerge --sync (or emaint -A sync, etc.) it will sync via git and emerge will work as always. Now if you want to do more or just want to learn more about git then that's different. Todd I think what we are talking about is viewing things like the changelogs and such, which are currently not synced with the tree. Or did we change to some other topic and I missed it? I tracked back to Alan Mackenzie's split of this thread . Dale That's fine. I was addressing Philip wanting to update via git. If you want more than just synching and emerge then learn the one or two git commands you need. As Rich said, just use git log and git diff if you want to see changes. It's not hard, just different. Todd Oh I see now. I was thinking he was talking about how to get the changelogs. It seems he was talking about the tree itself, I guess. Dale :-) :-)
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
Fernando Rodriguez wrote: On Saturday, August 29, 2015 10:48:16 AM Dale wrote: Todd Goodman wrote: * Philip Webb purs...@ca.inter.net [150828 18:35]: 150828 Rich Freeman wrote: To really appreciate git you should understand git objects and their references, what a commit, tree, and blob are. Also, the whole copy-on-write concept and content-hashing concept. I used to think git looked really complicated until I sat through a 1 hr talk that focused mostly on the data model. Once you understand the data model, you understand everything. That doesn't take a lot of time. It does take a moderate amount of time learning the right things. They're not found in the manpages. Like I said, beautiful design, horrible interface. So is there a Gentoo doc -- Wiki, presumably -- explaining to users -- users, not dev's or Git addicts -- the essentials of Git, so that they can readily update using it ? If so, I'm willing to see if I can use it ; if not, I would suggest it sb a top priority for dev's to write. You don't *need* to know anything about git to update using it. Just change your /etc/portage/repos.conf/gentoo.conf as Rich outlined (and move away your rsync'd /usr/portage or wherever your portage tree goes.) Then when you emerge --sync (or emaint -A sync, etc.) it will sync via git and emerge will work as always. Now if you want to do more or just want to learn more about git then that's different. Todd I think what we are talking about is viewing things like the changelogs and such, which are currently not synced with the tree. Or did we change to some other topic and I missed it? I tracked back to Alan Mackenzie's split of this thread . Dale :-) :-) It's probably easier to do this: # cd /usr/portage # rm -r * # git clone repo-uri . Then do the repos.conf changes. That way you don't have to worry about portage doing a shallow clone. If you already did it then just unshallow it as Rich pointed. Then to view the logs just: #cd /usr/portage/cat/pkg #git log . Then 'git show first few digits of commit hash' to view a commit diff. You can use git use dev-vcs/tig if you find it easier though I thought it was pretty useless so it only lasted about 10 secs. in my system. So basicly the only change is that instead of: # less ChangeLog (or whatever you use to read logs) You'll do: # git log . Actually, I use eix-sync to sync my tree. However I do it, I want it done within the usual setup and commands. Given the bumps we've already seen, I'm not wanting to change that just yet. Let the devs work out some of the kinks first. Oh, I use Kwrite to read the changelogs. If I'm stuck in a console, nano, head or cat works. Well, it did in the past anyway. May not now tho. Dale :-) :-)
[gentoo-user] Re: ncurses: reductio ad absurdum
Rich Freeman rich0 at gentoo.org writes: On Fri, Aug 28, 2015 at 9:19 AM, walt w41ter at gmail.com wrote: Note that /usr/portage/sys-libs/ncurses/Changelog was last updated on April 6, several months ago. Rhetorical question: what is the purpose of a Changelog? Gentoo is no longer maintaining the old Changelog files. The source of all change logs going forward is in git: I believe there is interest in creating the old-format Changelogs for the rsync servers. They won't be present in the git repository, since they're just redundant extra data to sync. Who last updated ncurses, and why? git whatchanged /usr/portage/sys-libs/ncurses /tmp/log.txt See the attachment. You can ask git for full diffs as well fairly easily, or show them for individual commits or whatever. Then github or gentoo's git viewer can show you it in a pretty picture. Hmmm. Changelogs are easy to view (less) and all quickly available on your system, should your issues with a particular ebuild coincide with network issues (cheap/sorry ISP for example). I use them extensively, now that I know of quite a few devs, read the gentoo-dev list and hack around the various ebuilds myself. It there an easy, automated way to have this inforation on my system(s) without extra keystrokes? I have become critically dependant on Changelogs to ferret out quite a few issues, so some suggestions on automating (via git) those Changelogs (not attached to the file name but certainly the concurrent data) would be keenly appreciated. I do not want to have to do what you suggest above, on a per package basis. I want all of the former Changelog data and automated in the background. So advise as to pathways or just waiting patiently as it is in somebody's todo list, would be greatly appreciated? Surely, I'm not asking too much? Explain? regards, James
[gentoo-user] Re: ncurses: reductio ad absurdum
Rich Freeman rich0 at gentoo.org writes: On Fri, Aug 28, 2015 at 10:24 AM, James wireless at tampabay.rr.com wrote: It there an easy, automated way to have this inforation on my system(s) without extra keystrokes? Either wait for somebody to provide the old-style changelogs (some consider this critical, some consider this pointless - I'll admit I tend to fall in the latter), or just sync from git and run git log/whatchanged path or even tig path. That really gets you the same info and more, and lets you filter by path, file, and so on. Well, my git (kung fu) is currently terse. So, how do I automate this so the data via Git is on my system, say synced once a day? What I do not want is a 'per package' effort; suggestions are most welcome as is syntactical examples. Also, if a commit touched many packages at once it will show up in the log for each of the packages, but looking at it will also show you all the other related changes. It really is a much more data-rich view. I agree with this. But since the Changlog data was a fundamental part of Gentoo, for a very long time, and the devs wisely chose to upgrade to git, I would think that this functionality would be provided via git, at least by some hacks or detailed example. I do not need another project and I do read and experiment quite a bit with git. But that growth is going to take time. In the meantime WE, the gentoo users should not be required to study git and roll our own solution, one at a time from previously well defined functionality, imho. Besides Git is quite a beast:: how long from inception to completion did the gentoo git migration take? years? a decade? While I do believe the ChangeLogs will show up again for those who prefer them, I think that taking a bit of time to learn to use git is going to make your life better in the long haul. I think that in time people will stop using ChangeLogs. But that's exactly the point; for Changelogs and such info, there should at least be a news item or dev-blog post on automating such things. Per packages solutions miss the point of convenience of Changelogs. Enhanced data beyond changelogs? Sure, but that caveat should not be a blocker to what was available. ymmv. The simple matter is Changelogs are very useful, quick and simple and once you see a seasoned dev did the work, most look elsewhere for problems. As you know there is quite a variance in dev skills, particularly with different languages and large/complex projects. wwr, James
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
On Fri, Aug 28, 2015 at 10:24 AM, James wirel...@tampabay.rr.com wrote: It there an easy, automated way to have this inforation on my system(s) without extra keystrokes? Either wait for somebody to provide the old-style changelogs (some consider this critical, some consider this pointless - I'll admit I tend to fall in the latter), or just sync from git and run git log/whatchanged path or even tig path. That really gets you the same info and more, and lets you filter by path, file, and so on. Also, if a commit touched many packages at once it will show up in the log for each of the packages, but looking at it will also show you all the other related changes. It really is a much more data-rich view. While I do believe the ChangeLogs will show up again for those who prefer them, I think that taking a bit of time to learn to use git is going to make your life better in the long haul. I think that in time people will stop using ChageLogs. -- Rich
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
Hello, Rich. On Fri, Aug 28, 2015 at 10:53:00AM -0400, Rich Freeman wrote: On Fri, Aug 28, 2015 at 10:24 AM, James wirel...@tampabay.rr.com wrote: It there an easy, automated way to have this inforation on my system(s) without extra keystrokes? [ ] While I do believe the ChangeLogs will show up again for those who prefer them, I think that taking a bit of time to learn to use git is going to make your life better in the long haul. I disagree completely. A little time spent on git is time wasted. Only a lot of time spent on git is useful. git is to VCSs as assembler is to programming languages. To use either effectively, you've got to have a complete grasp of the internal logic of git/the processor. This is in stark contrast with, say, Mercurial or CVS, or a language like C. For comparison, the collected man pages of git (collected into an info file) weigh in at 1.9 Mb. For Mercurial, the single man page is just 315k. I speak from bitter experience. I think that in time people will stop using ChangeLogs. I think people will be using ChangeLogs for as long as they exist. The ChangeLog is a very convenient and useful reference. By comparison, typing in arcane commands to git is a pain, even if you're only going to be doing it once whilst creating a script. -- Rich -- Alan Mackenzie (Nuremberg, Germany).
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
walt wrote: On Fri, 28 Aug 2015 15:29:20 -0400 Rich Freeman ri...@gentoo.org wrote: I used to think git looked really complicated until I sat through a one hour talk that focused mostly on the data model. Once you understand the data model, you understand everything. That doesn't take a lot of time. Does that talk happen to be available on youtube or some equivalent? I'm a git fanboi in spite of its failings. I've been using git since Linus and Larry McVoy divorced (amicably, claims Linus) over where the kernel source repo would reside. That amicable divorce spawned the development of git in the first place (says Linus). I'm no expert on git, but 'git bisect' has allowed me to file countless (for countless = than the number of cc's of vino in my glass) bug reports over the years. I see that all the gentooers who replied to my post have been lingering in this mailing list for years. (You are Old Farts, by my definition.) All gentoo Old Farts are here because we are gentoo addicts and not one of us could abandon gentoo even if we wanted to. Including me. That may help me too. It took me a while to have a sorta understanding of how Gentoo is set up ebuild wise and such and now I feel like I did back in 2003. Having it on youtube or something would be really good. I did find this tho: https://www.youtube.com/watch?v=73I5dRucCds I'll try to watch it later. Of course, that video leads to other videos since Youtube tries to show similar ones in its suggestion thingy. Dale :-) :-)
[gentoo-user] Re: ncurses: reductio ad absurdum
Rich Freeman rich0 at gentoo.org writes: https://bugs.gentoo.org/show_bug.cgi?id=558800 No big deal now that we can see all the recent changes to ncurses. Me; I'm just going to wait until Monday, to sync and update. The problem was that EAPI5 (which is only a year old or so) interacts with slot moves (which have been around for a long time) in a way that wasn't anticipated. Devs comfortable with slot moves didn't recognize that this would be a high-impact change, and so there wasn't much testing, and it directly hit stable. I think we'll be talking more about that from a lessons-learned perspective. I appreciate the explanation. I've already move on to other projects this weekend. I just hope there his a gentoo wiki page on GIT that minimizes the what we need to read and some clear examples on how to do things and why. I know it's evolving. Of course, ncurses being such an important package just made this a huge mess. Yep it first hit me a few days ago with libcaca and the LXQT install. All is fine now for me. PS. I use man pages as references. They rarely connect the dots like your previously referred to presentation. That would be keen to link off of a gentoo wiki page; or get one of the more knowledgable gentoo devs to provide something handy and simple. Lord knows, we gonna be revisiting GIT issues for a while. thx, James
[gentoo-user] Re: ncurses: reductio ad absurdum
walt w41ter at gmail.com writes: I avoided yesterday's downgrade from ncurses-6.0 to ncurses-5.9-r4 because it was obviously(?) a mistake. I believe you. But here's what I just experienced:: I syncd a few hours ago. Now when I just went to upgrade I got this gyration of the latest ncurses debacle:: [ebuild U #] sys-libs/ncurses-5.9-r5:0/5::gentoo [5.9-r4:0/0::gentoo] [ebuild NS #] sys-libs/ncurses-5.9-r99:5::gentoo [5.9-r4:0::gentoo] OK so I masked off those to in package.mask. I decided to up grade a few packages, one at a time. Upgrading openssl got this response:: Emerging (1 of 2) sys-libs/ncurses-6.0-r1::gentoo and it is not stable nor has it been unmasked. Something really funky/flunky be happening with ncurses (again). Now look:: [ebuild NS ~] sys-libs/ncurses-5.9-r101:5::gentoo [ebuild NS ~] sys-libs/ncurses-6.0:5/6::gentoo Surely I can mask off these updates and stay with :: sys-libs/ncurses- 5.9-r4 For a while, till things settle a bit? Weird. I mask off a version and a newer, later version appears. wtf? Who last updated ncurses, and why? I believe you have 'hit the nail' dead center. Maybe a systemd requirement perhaps? I do not know. I know, systemd is running git now? James
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
Rich Freeman ri...@gentoo.org wrote: On Fri, Aug 28, 2015 at 10:24 AM, James wirel...@tampabay.rr.com wrote: It there an easy, automated way to have this inforation on my system(s) without extra keystrokes? Either wait for somebody to provide the old-style changelogs (some consider this critical, some consider this pointless - I'll admit I tend to fall in the latter), or just sync from git and run git log/whatchanged path or even tig path. That really gets you the same info and more, and lets you filter by path, file, and so on. Also, if a commit touched many packages at once it will show up in the log for each of the packages, but looking at it will also show you all the other related changes. It really is a much more data-rich view. While I do believe the ChangeLogs will show up again for those who prefer them, I think that taking a bit of time to learn to use git is going to make your life better in the long haul. I think that in time people will stop using ChageLogs. So, do I need to change the rsync-type for gentoo o git, or how else do I do this -- get the changelogs on my actual box? Now, I don't see a .git under /usr/portage which makes sense, but is syncing done now by git? \ -- Your life is like a penny. You're going to lose it. The question is: How do you spend it? John Covici cov...@ccs.covici.com
[gentoo-user] Re: ncurses: reductio ad absurdum
Emanuele Rusconi emarsk at gmail.com writes: As a user - not specifically a Gentoo user - I only need to know 3 commands: When I experimented with managing my config files with git I did a lot of reading (I was new to VCS in general), and in the end I realized that, although git is really powerful and complex, for my needs I actually really needed to know just a handful of basic commands, and I could use tig and/or gitk for almost anything. (3) things really? You can see into the future now? I'm going to be as polite as I can and say that is 'head in the sand' naive. Read up on gitignore [1] and how the devs plan on using it. It was quite a discussion on gentoo-dev. Surely more is occurring 'back_channel' as it should to formulate a comprehensive scheme for just one aspect of git. All I suggested was basically the devs put a bit of extra thought into how gitignore would be used by users and overlays to be compatible with codes migrating from one's system to an overlay and maybe eventually the portage tree. Entertaining as the dev sarcasms were, it's was/is a valid concern because gitignore is used to prune/remove files in source trees. All users maintain (at least temporally a source tree. And in case you have not read up on it, we have something called 'epatch-user' where ordinary users can inject code to test and permanently modify packages. Sorry, but your position is 'clueless' even though well intentioned. We are all going to be learning quite a bit of 'git' like it or not. Git deployment is no excuse for crippling something as fundamental to gentoo as the changelogs. There are a myriad of valid reason that those changelogs are so pronounced and easy to find and explore. I'm an ordinary gentoo user (a gentoo_commoner), just so you know. wwr, James [1] http://git-scm.com/docs/gitignore
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
James, maybe you skimmed over the premise As a user - not specifically a Gentoo user? Should I explain its implications? I was specifically addressing the complaint that you need to be a git guru just to access the changelogs. You don't. As Rich Freeman already pointed out, it's really trivial, especially with tig. -- Emanuele Rusconi
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
On Fri, Aug 28, 2015 at 11:26 AM, James wirel...@tampabay.rr.com wrote: I agree with this. But since the Changlog data was a fundamental part of Gentoo, for a very long time, and the devs wisely chose to upgrade to git, I would think that this functionality would be provided via git, at least by some hacks or detailed example. See my email, and the follow-up with the repos.conf for fetching from git. I'm sure the docs will continue to evolve, and changelogs may be re-introduced in rsync. In the meantime WE, the gentoo users should not be required to study git and roll our own solution, one at a time from previously well defined functionality, imho. Besides Git is quite a beast:: how long from inception to completion did the gentoo git migration take? years? a decade? If you give it a few months I'm sure everything will settle out. We've taken WAY too long to complete the git migration, and most of the delay was in the interest of avoiding impacting users with change. We basically ended up drawing a line in the sand and saying that we knew there would be little issues like this that annoy some people to no end, and we'll just deal with that. Otherwise it would have been another 5 years before we end up cancelling the git migration and started talking about whatever is replacing git. But that's exactly the point; for Changelogs and such info, there should at least be a news item or dev-blog post on automating such things. Per packages solutions miss the point of convenience of Changelogs. Feel free to contribute a news item or blog post. :) And nobody is proposing any per-package solutions. Git covers the entire tree and we've standardized the commit messages/etc. Enhanced data beyond changelogs? Sure, but that caveat should not be a blocker to what was available. ymmv. Sure, and those old-format changelogs will probably exist again some day. If more people find them useful, they're more likely to get re-created. -- Rich
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
On Fri, Aug 28, 2015 at 12:22 PM, James wirel...@tampabay.rr.com wrote: Surely I can mask off these updates and stay with :: sys-libs/ncurses- 5.9-r4 For a while, till things settle a bit? Weird. I mask off a version and a newer, later version appears. wtf? In your long post you didn't actually say what version of ncurses you're expecting to have, or whether you're running stable or ~arch. I can't tell you EXACTLY what to do without knowing that. However, in general my advice would be to remove anything you added to any config files in the last few days that mentions ncurses, do an emerge --sync, and then update your system as usual, and it will probably work just fine. Masking off packages will probably make things worse. Who last updated ncurses, and why? I believe you have 'hit the nail' dead center. Maybe a systemd requirement perhaps? I do not know. I know, systemd is running git now? Well, I posted the git log earlier, but vapier made the initial commit. The intent was to adjust the package slot on ncurses to be more consistent with how everything else works, and it looks like this was done in part to make life easier on Gentoo Prefix: https://bugs.gentoo.org/show_bug.cgi?id=558800 The problem was that EAPI5 (which is only a year old or so) interacts with slot moves (which have been around for a long time) in a way that wasn't anticipated. Devs comfortable with slot moves didn't recognize that this would be a high-impact change, and so there wasn't much testing, and it directly hit stable. I think we'll be talking more about that from a lessons-learned perspective. Of course, ncurses being such an important package just made this a huge mess. -- Rich
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
Hello, Rich. On Fri, Aug 28, 2015 at 01:41:05PM -0400, Rich Freeman wrote: On Fri, Aug 28, 2015 at 11:26 AM, Alan Mackenzie a...@muc.de wrote: I disagree completely. A little time spent on git is time wasted. Only a lot of time spent on git is useful. I disagree with this. git is to VCSs as assembler is to programming languages. To use either effectively, you've got to have a complete grasp of the internal logic of git/the processor. I agree with this. the collected man pages of git (collected into an info file) weigh in at 1.9 Mb. And if you read them all 12 times it probably wouldn't help you out with using git one bit. I haven't read most of it, and I'd say that I grok it. You really need to take time to understand how git works, and that doesn't actually take that much time. I've spent many tens of hours in the last year trying to get to grips with git, ever since a project I work on (Emacs) converted to using git. There are all sorts of stupidities in it - like `push' and `pull' not being opposites, `clone' not producing a clone, but a new repository radically different from the original. Unnecessary arcane teminology, and standard terminology perversely used to mean something different. When everything works just fine, according to the book, I now have few problems. But when something goes wrong - like merge conflicts, for example, I end up spending, perhaps, two hours searching the doc for an answer then end up asking for help on the project mailing list. With the same problems in, say, hg or cvs, the answers would be clear within a few minutes of opening the doc. The man pages are mainly used to figure out what Linus named the command line option you're looking for when he was drunk and creating a new subcommand. Git has a beautiful design and a horrible interface. I can accept that. If you understand the design you'll know that there must be a way to do something, but you'll spend 15min figuring out which command line utility does it best. I don't want to have to understand the design. I just want to be a user. I've got enough things competing for limited mental capacity as it is. -- Rich -- Alan Mackenzie (Nuremberg, Germany).
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
On Fri, Aug 28, 2015 at 06:45:47PM +, Alan Mackenzie wrote: There are all sorts of stupidities in it - like `push' and `pull' not being opposites That's fair. `clone' not producing a clone, but a new repository radically different from the original. That is not true. It is a clone, just with all of the most recent versions of the files checked out and availble for editing. Unnecessary arcane teminology, and standard terminology perversely used to mean something different. That's also fair. I remember reading a bunch of stuff about the git team generally doing the exact opposite of Subversion whenever they could because they disliked it so much. If you understand the design you'll know that there must be a way to do something, but you'll spend 15min figuring out which command line utility does it best. I don't want to have to understand the design. I just want to be a user. I've got enough things competing for limited mental capacity as it is. What? I don't think Rich means understanding all of the implementation details, just high-level concepts like what a branch is, what a commit is, and merging and rebasing. These topics - branching, merging, etc. - are central to how a team works, so it is important to understand them and how whatever VCS you're using deals with them. Alec
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
150828 Rich Freeman wrote: To really appreciate git you should understand git objects and their references, what a commit, tree, and blob are. Also, the whole copy-on-write concept and content-hashing concept. I used to think git looked really complicated until I sat through a 1 hr talk that focused mostly on the data model. Once you understand the data model, you understand everything. That doesn't take a lot of time. It does take a moderate amount of time learning the right things. They're not found in the manpages. Like I said, beautiful design, horrible interface. So is there a Gentoo doc -- Wiki, presumably -- explaining to users -- users, not dev's or Git addicts -- the essentials of Git, so that they can readily update using it ? If so, I'm willing to see if I can use it ; if not, I would suggest it sb a top priority for dev's to write. -- ,, SUPPORT ___//___, Philip Webb ELECTRIC /] [] [] [] [] []| Cities Centre, University of Toronto TRANSIT`-O--O---' purslowatchassdotutorontodotca
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
On Fri, Aug 28, 2015 at 11:26 AM, Alan Mackenzie a...@muc.de wrote: I disagree completely. A little time spent on git is time wasted. Only a lot of time spent on git is useful. I disagree with this. git is to VCSs as assembler is to programming languages. To use either effectively, you've got to have a complete grasp of the internal logic of git/the processor. I agree with this. the collected man pages of git (collected into an info file) weigh in at 1.9 Mb. And if you read them all 12 times it probably wouldn't help you out with using git one bit. I haven't read most of it, and I'd say that I grok it. You really need to take time to understand how git works, and that doesn't actually take that much time. The man pages are mainly used to figure out what Linus named the command line option you're looking for when he was drunk and creating a new subcommand. Git has a beautiful design and a horrible interface. If you understand the design you'll know that there must be a way to do something, but you'll spend 15min figuring out which command line utility does it best. -- Rich
Re: [gentoo-user] Re: ncurses: reductio ad absurdum
On Fri, Aug 28, 2015 at 3:10 PM, Alec Ten Harmsel a...@alectenharmsel.com wrote: On Fri, Aug 28, 2015 at 06:45:47PM +, Alan Mackenzie wrote: I don't want to have to understand the design. I just want to be a user. I've got enough things competing for limited mental capacity as it is. What? I don't think Rich means understanding all of the implementation details, just high-level concepts like what a branch is, what a commit is, and merging and rebasing. These topics - branching, merging, etc. - are central to how a team works, so it is important to understand them and how whatever VCS you're using deals with them. Well, I'd go a bit further. To really appreciate git you should understand git objects and their references, what a commit, tree, and blob are. Also, the whole copy-on-write concept and content-hashing concept. I used to think git looked really complicated until I sat through a one hour talk that focused mostly on the data model. Once you understand the data model, you understand everything. That doesn't take a lot of time. It does take a moderate amount of time spent learning the right things. They're not found in the manpages. I'm not disagreeing at all with many of the gripes about git. I still share most of them, but now I feel like I'm the one in control and the fact that git pull doesn't rebase by default is just an annoyance and not a source of arcane behavior. Like I said, beautiful design, horrible interface. -- Rich
[gentoo-user] Re: ncurses: reductio ad absurdum
On Fri, 28 Aug 2015 15:29:20 -0400 Rich Freeman ri...@gentoo.org wrote: I used to think git looked really complicated until I sat through a one hour talk that focused mostly on the data model. Once you understand the data model, you understand everything. That doesn't take a lot of time. Does that talk happen to be available on youtube or some equivalent? I'm a git fanboi in spite of its failings. I've been using git since Linus and Larry McVoy divorced (amicably, claims Linus) over where the kernel source repo would reside. That amicable divorce spawned the development of git in the first place (says Linus). I'm no expert on git, but 'git bisect' has allowed me to file countless (for countless = than the number of cc's of vino in my glass) bug reports over the years. I see that all the gentooers who replied to my post have been lingering in this mailing list for years. (You are Old Farts, by my definition.) All gentoo Old Farts are here because we are gentoo addicts and not one of us could abandon gentoo even if we wanted to. Including me.