Am 17.04.2009 um 19:44 schrieb D. Richard Hipp: (...)
>> Copying a Working Directory >> --------------------------- >> I'm a little confused about what happens when I make a copy of the >> directory containing my repository and the source code. They seem to >> stay linked somehow. Is this a bug? Is it a feature? What's the >> proper fossil technique? (...) > It is a feature. In hg, the working directory and the repository all > all wrapped together as one. With fossil they are distinct. If you > want a separate copy of the repository, you'll need to clone it using > the "fossil clone" command. Then check out the new working directory > from the clone. Also: be sure to turn off autosync in the clone or > it will try to automatically push its changes back to the original. Perhaps I did get it wrong, but I thought, he had written: (...) when I make a copy of the directory containing my repository *and* the source code. (...) Emphasis mine! So I thought he had copied the repository as well! So now he was astonished why both repositories seem to be connected somehow ... If I did not understood him wrong, than it's worth mentioning, that those both *repositories* are *not* connected together! The checkouts are, however. Suppose he has a directory like this: /tmp/XXX/src /tmp/XXX/_FOSSIL_ /tmp/XXX/the_repository.fsl then the file '/tmp/XXX/_FOSSIL_' contains a reference to '/tmp/XXX/ the_repository.fsl'. If he now copies 'XXX' to 'YYY' he will get a structure like: /tmp/YYY/src /tmp/YYY/_FOSSIL_ /tmp/YYY/the_repository.fsl But '/tmp/YYY/_FOSSIL_' still contains a reference to '/tmp/XXX/ the_repository.fsl' as the copy was not performed via fossil but some third-party tool like e.g. 'cp'. So fossil does not know about that copying took place. If he now switch into '/tmp/YYY' and commit some changes, they will be committed into '/tmp/XXX/the_repository.fsl' and not the one he thought!!! Bad trap for sure, as the checkout of '/tmp/YYY' is indeed decoupled from the repository '/tmp/YYY/the_repository.fsl'! So if he still wants to work that way, he should either 'fossil clone' that repository, as you proposed, or he should 'fossil close; fossil open' within '/tmp/YYY' right after the copy. I feel, the cloning is the fossil way to go! However, if I did understood him wrong and he did, what you all already answered to him, please sorry the noise I made ... Ciao, chi :-) _______________________________________________ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
