[fossil-users] I have two trunks?
Hi list; Today I opened one of my repos, and expected to see some stuff that is in 'trunk'. However, it was empty as the day it was born. I could see all my branches via 'branches', and could update to them, but when I would update to 'trunk' it would always be empty. Odd, I thought, I'm quite sure I put stuff there, but 'oh, well', I thought, and I checked some stuff into it that I had expected to already be there, and then also made a branch for some other things, and checked them in, too. When I look at the timeline via fossil ui, I very clearly see I now have two trunks, the original one, and this new one. I didn't even think this was possible, because I believed that fossil repos only have one unified DAG. Am I insane (er, 'am I in error', I mean to ask), or is this normal, and due to something odd that I did (though I can't imagine what it would be). This is not a secret repo; you can look here if curious: fossil clone http://chiselapp.com/user/ziggurat29/repository/z29-3rdParty 3rdParty.fossil Thanks! -dave ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] I have two trunks?
Hi Dave. This is a fork (unintentional branch). It would happen (for example) if two different clones of a repository are updated independently of each other, and then sync'd. You can merge the errant trunk back to the good trunk without ill effect (minding you might have to do conflict resolution). If you think of the case of two different clones being updated as I described above, then ultimately merged together (creating a fork in a branch), you'll see the DAG is maintained. The labels for the branchname are maintained too, and the effect you see (two Trunk branches) is consistent w/ the model. Re-merge them and keep on developing ! -bch On 10/2/14, dave d...@ziggurat29.com wrote: Hi list; Today I opened one of my repos, and expected to see some stuff that is in 'trunk'. However, it was empty as the day it was born. I could see all my branches via 'branches', and could update to them, but when I would update to 'trunk' it would always be empty. Odd, I thought, I'm quite sure I put stuff there, but 'oh, well', I thought, and I checked some stuff into it that I had expected to already be there, and then also made a branch for some other things, and checked them in, too. When I look at the timeline via fossil ui, I very clearly see I now have two trunks, the original one, and this new one. I didn't even think this was possible, because I believed that fossil repos only have one unified DAG. Am I insane (er, 'am I in error', I mean to ask), or is this normal, and due to something odd that I did (though I can't imagine what it would be). This is not a secret repo; you can look here if curious: fossil clone http://chiselapp.com/user/ziggurat29/repository/z29-3rdParty 3rdParty.fossil Thanks! -dave ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] I have two trunks?
On Thu, Oct 2, 2014 at 7:29 PM, B Harder brad.har...@gmail.com wrote: This is a fork (unintentional branch). It would happen (for example) if two different clones of a repository are updated independently of each other, and then sync'd. You can merge the errant trunk back to The weird thing is, he's got two initial empty commits. i'm at a loss to explain that. the good trunk without ill effect (minding you might have to do conflict resolution). If you think of the case of two different clones being updated as I described above, then ultimately merged together (creating a fork in a branch), you'll see the DAG is maintained. The labels for the branchname are maintained too, and the effect you see (two Trunk branches) is consistent w/ the model. Re-merge them and keep on developing ! If both clones were detached from the start, that could explain it, it guess. http://chiselapp.com/user/ziggurat29/repository/z29-3rdParty/timeline -- - stephan beal http://wanderinghorse.net/home/stephan/ http://gplus.to/sgbeal Freedom is sloppy. But since tyranny's the only guaranteed byproduct of those who insist on a perfect world, freedom will have to do. -- Bigby Wolf ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] I have two trunks?
Thus said Stephan Beal on Thu, 02 Oct 2014 19:31:12 +0200: The weird thing is, he's got two initial empty commits. i'm at a loss to explain that. Seems that one of them must have been created with an older fossil and the other created with a newer version from Fossil's trunk which currently does not force an initial commit comment of ``initial empty check-in'' when the fossil is created. Would be interesting to know how the repository was created. I don't know how Chiselapp imports a repository, but is it possible that the it might have created the additional trunk when it was uploaded? Andy -- TAI64 timestamp: 4000542d9630 ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] I have two trunks?
On Oct 2, 2014 11:14 AM, Andy Bradford amb-fos...@bradfords.org wrote: Thus said Stephan Beal on Thu, 02 Oct 2014 19:31:12 +0200: The weird thing is, he's got two initial empty commits. i'm at a loss to explain that. Seems that one of them must have been created with an older fossil and the other created with a newer version from Fossil's trunk which currently does not force an initial commit comment of ``initial empty check-in'' when the fossil is created. Is that even possible? I thought the repo would have to be created once (and only once), generating it's repo-id, and then cloned for all subsequent copies before things can begin deviating. Would be interesting to know how the repository was created. I don't know how Chiselapp imports a repository, but is it possible that the it might have created the additional trunk when it was uploaded? Andy -- TAI64 timestamp: 4000542d9630 ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] I have two trunks?
Thing is (well, two things): a) I would expect a fork to still have a common ancestor. I have certainly run across the fork thing due to out of synced repos (and as I recall, I get a warning beforehand). And b) in this case I had started from a freshly pulled-down clone, so I never had the opportunity to be out-of-sync. Ultimately, it's not a crisis for me. This repo is a 'utility' repo that doesn't have code or precious stuff in it. I mention it partially as a curiousity, and in case it is symptomatic of a bug, since I didn't think it was possible to have two trees with no common ancestor (at least, the way it is is depicted in the gui), and certainly not through the way I got here. As fate would have it, I'm going to destroy this repo and rebuild it anyway, but I might keep a copy on-hand for later scrutiny in my copious free time -Original Message- From: fossil-users-boun...@lists.fossil-scm.org [mailto:fossil-users-boun...@lists.fossil-scm.org] On Behalf Of B Harder Sent: Thursday, October 02, 2014 12:29 PM To: Fossil SCM user's discussion Subject: Re: [fossil-users] I have two trunks? Hi Dave. This is a fork (unintentional branch). It would happen (for example) if two different clones of a repository are updated independently of each other, and then sync'd. You can merge the errant trunk back to the good trunk without ill effect (minding you might have to do conflict resolution). If you think of the case of two different clones being updated as I described above, then ultimately merged together (creating a fork in a branch), you'll see the DAG is maintained. The labels for the branchname are maintained too, and the effect you see (two Trunk branches) is consistent w/ the model. Re-merge them and keep on developing ! -bch On 10/2/14, dave d...@ziggurat29.com wrote: Hi list; Today I opened one of my repos, and expected to see some stuff that is in 'trunk'. However, it was empty as the day it was born. I could see all my branches via 'branches', and could update to them, but when I would update to 'trunk' it would always be empty. Odd, I thought, I'm quite sure I put stuff there, but 'oh, well', I thought, and I checked some stuff into it that I had expected to already be there, and then also made a branch for some other things, and checked them in, too. When I look at the timeline via fossil ui, I very clearly see I now have two trunks, the original one, and this new one. I didn't even think this was possible, because I believed that fossil repos only have one unified DAG. Am I insane (er, 'am I in error', I mean to ask), or is this normal, and due to something odd that I did (though I can't imagine what it would be). This is not a secret repo; you can look here if curious: fossil clone http://chiselapp.com/user/ziggurat29/repository/z29-3rdParty 3rdParty.fossil Thanks! -dave ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/foss il-users ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] I have two trunks?
-Original Message- From: fossil-users-boun...@lists.fossil-scm.org [mailto:fossil-users-boun...@lists.fossil-scm.org] On Behalf Of Andy Bradford Sent: Thursday, October 02, 2014 1:15 PM To: Stephan Beal Cc: Fossil SCM user's discussion Subject: Re: [fossil-users] I have two trunks? Thus said Stephan Beal on Thu, 02 Oct 2014 19:31:12 +0200: The weird thing is, he's got two initial empty commits. i'm at a loss to explain that. Seems that one of them must have been created with an older fossil and the other created with a newer version from Fossil's trunk which currently does not force an initial commit comment of ``initial empty check-in'' when the fossil is created. Would be interesting to know how the repository was created. I don't know how Chiselapp imports a repository, but is it possible that the it might have created the additional trunk when it was uploaded? Andy I'm tending to suspect not (additional trunk at creation) for a reason that I will soon make clear, but the 'version mismatch' hypothesis is interesting. Fyi, the creation process went like this: * I created my repo locally, with current (v1.29, June 12th) fossil binary. I added a few things to trunk, and made some branches and added a bunch of stuff there. I decided to put it on a third party hosting site so I could avoid hosting myself. * Its big, so chisel can't import it, but they have a feature where you explicitly supply the project id, and then you can sync your big repo into it. So I did that. * I tossed my local, and then cloned down the one on chisel. (Have done this in several places, actually.) * I later added a new branch successfully as expected off trunk (in this case named 'sqlite3-3.8.6'). This is why I don't think the 'new trunk upon upload' theory is true (assuming my recollection is correct) * Then today, as mentioned, I cloned afresh to a different machine (which is why I disbelieve the 'fork because of out-of-sync repos' hypothesis), and opened, and this is when I noticed I had the apparently empty root. I did not check the gui at this time, alas, I figured I was getting old. So I checked stuff into root that I was pretty darn sure was already there (because if it wasn't it would have made creating all those other branches painful, and I would have remembered that). * I take a look at the timeline on the gui, and Lo! And Behold! What appears to be multiple trunks. Anyway, again, not a crisis for me, but I didn't think this was possible in the schema of things, and thought somebody might like to know if it is a symptom of some sort of bug. -dave ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] I have two trunks?
On 10/2/2014 10:31 AM, Stephan Beal wrote: The weird thing is, he's got two initial empty commits. i'm at a loss to explain that. http://chiselapp.com/user/ziggurat29/repository/z29-3rdParty/timeline I think I can partly explain the two initial empty commits part, or at least show off a similarly strange effect. I recently moved a private repository to Chiselapp so I could point at it from a series of blog posts. (Shameless plug: http://curiouser.cheshireeng.com/2014/09/11/wp-cli-utilities-source-code/ :-) I had trouble with getting it to upload my repo directly, so I chose to let it create a fresh repo with the project code set to match mine, which is a feature chiselapp publicly offers. The result is a timeline where there are two initial empty commits, and amusingly since I haven't added anything to it since I published it, the second one is *after* all of the actual work. I suppose I should wonder what happens if a fresh clone is made... http://chiselapp.com/user/rberteig/repository/WPCLI/timeline I think this is an artifact of creating an empty repo, changing its project ID, and then syncing a bunch of older changes. I'm not sure if it is harmful or just confusing. Since it is now possible (IIRC) to create a repo without the initial empty commit, perhaps chiselapp should do that when it knows it is create a repo with a specific project ID. And, actually, the clone experiment *is* more interesting than I expected. I just cloned from my WPCLI and opened it in an empty folder. The result is a folder containing only the _FOSSIL_ file. It opened [9f3230], aka the most recent checkin tagged trunk, which is the second empty checkin. I tried fossil merge 3b301fad10 and it failed because there is no common ancestor. fossil update 3b301fad10 works as expected. One easy answer should be to move the extra initial checkin to the mistake branch and close it. However, attempting to do that reveals something else odd: the info page for the checkin shows the raw artifact and not the usual set of useful links including edit. https://chiselapp.com/user/rberteig/repository/WPCLI/info/9f323048ac Is this a chiselapp bug or a fossil bug? I don't know, but I'll likely poke at it a little more after lunch. -- Ross Berteig r...@cheshireeng.com Cheshire Engineering Corp. http://www.CheshireEng.com/ ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] I have two trunks?
Bingo. On 10/2/2014 11:33 AM, dave wrote: I'm tending to suspect not (additional trunk at creation) for a reason that I will soon make clear, but the 'version mismatch' hypothesis is interesting. Fyi, the creation process went like this: * I created my repo locally, with current (v1.29, June 12th) fossil binary. I added a few things to trunk, and made some branches and added a bunch of stuff there. I decided to put it on a third party hosting site so I could avoid hosting myself. * Its big, so chisel can't import it, but they have a feature where you explicitly supply the project id, and then you can sync your big repo into it. So I did that. I'm currently running a hair older fossil than you, but very similar vintage and likely newer than the one on chiselapp. version 1.29 [77f53423ae] I followed essentially the same path, except my repo was small enough that it should have allowed the upload but failed. I wonder if a version mismatch was part of that story? It certainly could be. Chiselapp's shows Fossil version [3d49f04587] 2014-01-27 17:33:44 as the title text for the link back to fossil-scm.org in the page footer in their default skin. Going by the date alone, it is *older*. -- Ross Berteig r...@cheshireeng.com Cheshire Engineering Corp. http://www.CheshireEng.com/ ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] I have two trunks?
Thus said B Harder on Thu, 02 Oct 2014 11:17:23 -0700: Is that even possible? I thought the repo would have to be created once (and only once), generating it's repo-id, and then cloned for all subsequent copies before things can begin deviating. The project-id is stored in a table in a database---it can be changed. I suspect that this is how Chiselapp actually allows you to sync a new repository to their site in some configurations. Because the act of creating a new repository necessarily also includes an initial commit (older versions of Fossil), you end up with two trunks because there were actually 2 repositories created, not one. The ability to create a repository without an initial commit might be something that Chiselapp could take advantage of in this case. Andy -- TAI64 timestamp: 4000542d9f96 ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] I have two trunks?
... Thus said B Harder on Thu, 02 Oct 2014 11:17:23 -0700: Is that even possible? I thought the repo would have to be created once (and only once), generating it's repo-id, and then cloned for all subsequent copies before things can begin deviating. The project-id is stored in a table in a database---it can be changed. I suspect that this is how Chiselapp actually allows you to sync a new repository to their site in some configurations. Because the act of creating a new repository necessarily also includes an initial commit (older versions of Fossil), you end up with two trunks because there were actually 2 repositories created, not one. The ability to create a repository without an initial commit might be something that Chiselapp could take advantage of in this case. ... Yes, I think you guys solved the mystery (side effect of the project ID change magickry). Now the question is whether that behaviour of creating two disjoint trees in one repo upon project id change is a bug. Well, I'll leave that for others to ponder; I'm satisfied for now, and I'll make a concientious effort in the the future to merge the two trunks immediately to prevent downstream weirdness as a coping strategy. Thanks- -dave ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] I have two trunks?
2014-10-02 20:54 GMT+02:00 Andy Bradford: The project-id is stored in a table in a database---it can be changed. I suspect that this is how Chiselapp actually allows you to sync a new repository to their site in some configurations. Because the act of creating a new repository necessarily also includes an initial commit (older versions of Fossil), you end up with two trunks because there were actually 2 repositories created, not one. The ability to create a repository without an initial commit might be something that Chiselapp could take advantage of in this case. Nice explanation! Thanks! 2014-10-02 21:02 GMT+02:00 dave d...@ziggurat29.com: Now the question is whether that behaviour of creating two disjoint trees in one repo upon project id change is a bug. Yes, this sounds like a bug in chiselapp (not in fossil!). Normally, synchronising two separate repositories (which each have their own initial empty checkin) is not possible because they have a different project-id. But if chiselapp allows to change the project-id, then the two different initial check-ins will be moved over to both sides of the sync. fossil new --empty would indeed help chiselapp here (which is already available in fossil 1.29), preventing the two initial checkins in this situation. Regards, Jan Nijtmans ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] I have two trunks?
If it helps: I tried something very simple (in this example: https://chiselapp.com/user/rberteig/repository/WPCLI/home), and it apparently fixed the problem in cloned copy at least. I shunned the empty initial checking, rebuilt, and re-enabled the artifact (just in case). The timeline shows it removed! ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] I have two trunks?
-Original Message- From: fossil-users-boun...@lists.fossil-scm.org [mailto:fossil-users-boun...@lists.fossil-scm.org] On Behalf Of to...@acm.org Sent: Thursday, October 02, 2014 2:58 PM To: Fossil SCM user's discussion Subject: Re: [fossil-users] I have two trunks? If it helps: I tried something very simple (in this example: https://chiselapp.com/user/rberteig/repository/WPCLI/home), and it apparently fixed the problem in cloned copy at least. I shunned the empty initial checking, rebuilt, and re-enabled the artifact (just in case). The timeline shows it removed! It does help. I'll make shunning the spurious checkin (along with the frumious Bandersnatch) part of my workflow when 'importing' to chiselapps in this way. -dave ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] I have two trunks?
On Thu, Oct 2, 2014 at 2:33 PM, dave d...@ziggurat29.com wrote: * Its big, so chisel can't import it, but they have a feature where you explicitly supply the project id, and then you can sync your big repo into it. So I did that. When I did have a repo on Chisel, I didn't notice this feature. So, what I did was to create a new repo on Chisel, clone to my local PC, then imported from the original repo to the local clone of the Chisel repo, then did a sync. If there was anything strange, I never noticed it. ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] I have two trunks?
-Original Message- From: fossil-users-boun...@lists.fossil-scm.org [mailto:fossil-users-boun...@lists.fossil-scm.org] On Behalf Of Ron W Sent: Thursday, October 02, 2014 3:35 PM To: Fossil SCM user's discussion Subject: Re: [fossil-users] I have two trunks? On Thu, Oct 2, 2014 at 2:33 PM, dave d...@ziggurat29.com wrote: * Its big, so chisel can't import it, but they have a feature where you explicitly supply the project id, and then you can sync your big repo into it. So I did that. When I did have a repo on Chisel, I didn't notice this feature. So, what I did was to create a new repo on Chisel, clone to my local PC, then imported from the original repo to the local clone of the Chisel repo, then did a sync. If there was anything strange, I never noticed it. Oh, fascinating, I hadn't considered using the export/import mechanism to transfer it over (if I understand you correctly). I guess I would lose things like wikis tickets whatnot doing that, but many times this would not be material. ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] I have two trunks?
On 10/2/2014 12:57 PM, to...@acm.org wrote: I tried something very simple (in this example: https://chiselapp.com/user/rberteig/repository/WPCLI/home), and it apparently fixed the problem in cloned copy at least. I shunned the empty initial checking, rebuilt, and re-enabled the artifact (just in case). The timeline shows it removed! Cool. What I just did is manually enter the Edit link it won't show me, which would be http://chiselapp.com/.../ci_edit?r=9f323048ac;. When that gave me the checkin editor for that commit, I was able to move it to a different branch, close the leaf, and change the comment. I've confirmed from a fresh clone and open that the result is a much better experience for the poor user who is just trying to see the code. My polite request to the maintainer of chisellapp.com would be to try the (new-ish) feature of creating a new repository with no initial checkin. -- Ross Berteig r...@cheshireeng.com Cheshire Engineering Corp. http://www.CheshireEng.com/ ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
