Re: "tla build-config" question and suggestion
Pedro Perez <[email protected]>
| Newsgroups | gmane.comp.version-control.arch.user |
|---|---|
| Message-ID | <[email protected]> |
Andy Tai wrote: >On 4/12/06, Thomas Lord <[email protected]> wrote: > > >>Andy Tai wrote: >> >> >>>On 4/12/06, Pedro Perez <[email protected]> wrote: >>> >>> >>> >>> >>>>Apparently the extra dot in front of ./static is not an acceptable >>>>character/path for tla. >>>> >>>> >>>> >>>> >>>This is a weakness in the current code and shall be fixed. (added to >>>the to do list...) >>> >>> >>> >>Sigh. It's a security feature. Note that it is there deliberately >>(see the >>call to `is_non_upwards_relative_path'). I strongly recommend that you >>NOT CHANGE THAT BEHAVIOR. >> >> >> > >OK, once Tom has spoken, that's it. I will follow Tom's suggestion >here. Pedro, please just use the path without the dot in front. > > That's why I enjoy being a member of this list. I always learn something new. Thanks. ( I will check the code for is_non_upward_relative_path ). > > >>It would be a piss-poor feature of a revision control system if, just >>checking >>out a tree in the usual way without carefully scrutinizing the config file >>first, stuff could be installed where you don't expect >> >> > > > >>>>So question one is, is a double dot (parent directory) an illegal path >>>>for "tla build-config" ? >>>>Question two is, why ./lfs/. does not work but ./lfs does work? >>>> >>>> >>>> >>>> >>>OK, this is a case that is nice to be handled correctly as well. (a >>>low-priority to do) >>> >>> >>> >>> >>That's generous of you. I wouldn't bother. Why add code (i.e., >>additional >>sources of error, additional maintenance burden) just to give users the >>option of typing a gratuitous no-op? >> >>-t >> >> > > > >As suggested by Tom as well. > >I will still go ahead and implement the check-config suggestion by >Pedro, which is useful to have. > > >-- >Andy Tai, [email protected] > > > Thanks Tom, Andy, -- Pedro Perez