"We emphatically do NOT want Fossil second-guessing what branch you want to be on when you do "fossil update" without an argument."
Whatever option you decide - either way you are second guessing the users intent. 1. Current behavior is really, really confusing. 2. You *were* on trunk but magically ended up on some other branch. 3. It has caused real-world wasted time for me and for others that I support. The right thing from a user perspective is to either WARN the user that the branch was swizzled out from under them or WARN them and update to the actual branch you were on to start with, which is what the user is (rightly or wrongly) expecting. Silently changing branches without a loud warning does not make sense. A branch is a symbolic idea and our situational awareness is attached to the branch name, not to some abstract and non-intuitive notion of location regarding attached nodes on a DAG. In short a very loud warning is needed, no matter what path you take. Just my $0.02. On Fri, May 29, 2015 at 9:05 AM, j. van den hoff <veedeeh...@googlemail.com> wrote: > On Fri, 29 May 2015 17:38:39 +0200, Richard Hipp <d...@sqlite.org> wrote: > > On 5/29/15, Matt Welland <mattrwell...@gmail.com> wrote: >> >>> This is an exceedingly confusing behavior from fossil but the fix is >>> easy. >>> Just do "fossil up trunk". >>> >>> >> Indeed - Fossil is doing exactly the right thing here. If you just >> "fossil update" it advances you to the tip of whatever branch your are >> currently on. If you want to be at the tip of trunk, you really do >> need to say "fossil update trunk". >> >> We emphatically do NOT want Fossil second-guessing what branch you >> want to be on when you do "fossil update" without an argument. >> >> > for sure not. OTOH: although it has not (yet) happened to me, I can easily > see that it is confusing to leave trunk if I *am* on trunk before the > update and just issue `fossil up' (which most of the time for most of the > users seems to mean "move me forward on this branch (especially trunk) to > the current, last "version" of this branch" rather than "move me to > wherever my current checkout has evolved". at least that would be me > guess...). so while fossil should not second-guess my intention and > implement it, it very well _might_ second-guess my intention and notify me > when I am probably doing something unintentional. > > as with assorted other things (e.g. empty checkin messages), should fossil > not ask at this point (with a default of staying on the current branch)? > the scenario would be > > * I am on branch X > * I issue `fossil up' w/o argument > * `fossil' notes that this will move me to a different branch since > someone else has transplanted the last checkin to a different branch > * fossil asks if I do want to proceed or simply stay where I am. > > it should happen seldom enough in order not to annoy anybody if such a > question is issued. > > thx/j > > > -- > Using Opera's revolutionary email client: http://www.opera.com/mail/ > > _______________________________________________ > 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
