On Tue, Jul 22, 2014 at 11:35 AM, Michai Ramakers <[email protected]>
wrote:

> (FWIW, I know there have been flaws in this area before, and although
> I am not a dev having/willing to deal with them, I agree with Jan that
>

AFAIR i reverted that to not be the default behaviour (IMO it should not be
because it's a long-standing historical behaviour and we don't fully know
the implications of removing it).


> results in a cleaner design. I don't know how well initial-commit-less
> repos work with older fossil-versions, but I assume that has been
> discussed or considered in detail elsewhere. </EUR0.02>)
>

LOL! Late last year Jan and i discussed it a bit (not at length) after i
started experimenting with it in libfossil and was curious if fossil could
support it. Jan was the only one brave/gutsy enough to try it out! i knew
from having ported over so many algos from fossil that there were numerous
corner cases, assumptions, and assertions that expected the initial RID to
be in place, but those were just "historical coding decisions" - nothing in
the architecture specifically (or implicitly) prohibits an empty repo. The
bugs we've seen since introducing empty repos have been patching up former
assumptions which are not true in the empty-repo case.


> ---( without initial commit: )---
>
> ...

/tmp/f/a$ f up
> /tmp/f/a$ f tim
> === 2014-07-22 ===
> 09:20:06 [c3450bfdff] test (user: michai tags: trunk)
> +++ no more data (1) +++
> /tmp/f/a$ ls
> /tmp/f/a$
>
> ---( with forced initial commit: )---
>
> Tue Jul 22 11:22:04 CEST 2014
>

Same time zone :)


> ...

... "fossil undo" is available to undo changes to the working checkout.
> /tmp/f/a$ f tim
> === 2014-07-22 ===
> 09:22:41 [4b720ef551] *CURRENT* test (user: michai tags: trunk)
> === 2014-01-01 ===
> 00:00:00 [b58cc4d981] initial empty check-in (user: michai tags: trunk)
> +++ no more data (2) +++
>

(That last line is a little confusing, IMO. Almost looks like an error.)


> The order of opening/committing are important; if the commit from
> workdir 'b' happens before the repo is opened in workdir 'a',
> everything seems to work OK.
>
> In the 1st session (without initial empty commit), more commits can be
> done from dir 'b', but these are not seen when updating in dir 'a'
> afterwards. The timeline in workdir 'a' does show the commits.
>

i think i'm going to wait for Andy's response, rather than try to wrap my
head around that ;).

-- 
----- 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
[email protected]
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Reply via email to