Re: [fossil-users] Couple of newbie questions on artifacts and comparing
On Tue, 08 Jan, Stephan Beal wrote: > On Tue, Jan 8, 2013 at 3:49 PM, Stefan Bellon > wrote: > > > Is subversion equally frowned upon? If not, what about: > > > > fossil diff --from PREV --to HEAD myfile.c > > > > That would be too easy ;). Yes, it covers the most common case, but it > would be much more interesting to be able to say "go back N > versions", as such a feature implies the "prev" behaviour as well. fossil diff --from -5 --to -3 myfile.c it is, then. ;-) -0 (or 0) being the current and positive (non-zero) values making no sense (except when being run from inside the time machine shell). > > The question is whether PREV and HEAD (and perhaps other symbolic > > names) should be bound to one branch (i.e. the one of the current > > checked out tree) or whether it should be global and possibly cross > > branches. > > That's a good point. My gut feeling is that the currently-checked-out > branch "should" be used, defaulting to trunk if there is no checkout > (that said, i think we once had a thread where the user had removed > or renamed his trunk branch to something else). Or maybe a syntax like > branch_name:symbolic_name (which of course only works as long as > nobody uses ':' in their branch names). No need for a new syntax (and a very non-fossil one), I think. "fossil diff" already has the option --branch. Currently it is mutually exclusive with --from and --to (because the revision hashes already specify the branch), but that could be used to specify another branch than the current one when using --from and --to with symbolic (or commit-relative) names. Greetings, Stefan -- Stefan Bellon ___ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Couple of newbie questions on artifacts and comparing
On Tue, Jan 08, 2013 at 03:56:15PM +0100, Stephan Beal wrote: >On Tue, Jan 8, 2013 at 3:49 PM, Stefan Bellon wrote: > > Is subversion equally frowned upon? If not, what about: >fossil diff --from PREV --to HEAD myfile.c > >That would be too easy ;). Yes, it covers the most common case, but it >would be much more interesting to be able to say "go back N versions", as >such a feature implies the "prev" behaviour as well. > > > The question is whether PREV and HEAD (and perhaps other symbolic names) > should be bound to one branch (i.e. the one of the current checked out > tree) or whether it should be global and possibly cross branches. > >That's a good point. My gut feeling is that the currently-checked-out >branch "should" be used, defaulting to trunk if there is no checkout (that >said, i think we once had a thread where the user had removed or renamed >his trunk branch to something else). Or maybe a syntax like >branch_name:symbolic_name (which of course only works as long as nobody >uses ':' in their branch names). Actually, the "prev" keyword already does something.. Example: $ fossil diff --from prev will give you diff between current checkout from previous version on repository. or $ fossil diff --from prev myfile.c Here, prev mean previous version globally, not previous version of the specified file... This is not on the help, but I found this on the source.. http://fossil-scm.org/index.html/artifact/ab3b8161d773d9619a709c92ef33b799cc473a74?ln=49-72 -- Martin G. ___ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Couple of newbie questions on artifacts and comparing
On Tue, Jan 8, 2013 at 9:49 AM, Stefan Bellon wrote: > > Is subversion equally frowned upon? If not, what about: > > fossil diff --from PREV --to HEAD myfile.c > Fossil already does this, with the exception that "PREV" is called "prev" or "previous" and "HEAD" is called "current". fossil diff --from prev --to current myfile.c When you say --to current, that means use the version of the file that is checked in. If you omit the --to current, that means use the version of the file as editted in the check-out. In place of "prev" and "current" you can also use "tip" to mean the latest check-in. Or you can put the name of any branch, and that means the latest check-in on that branch. Or you can do things like "branch:2012-09-16" to mean the latest checkin on "branch" on or before 2012-09-16 00:00:00. See http://www.fossil-scm.org/fossil/doc/trunk/www/checkin_names.wiki for additional information and for more examples. -- D. Richard Hipp [email protected] ___ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Couple of newbie questions on artifacts and comparing
On Tue, Jan 8, 2013 at 3:49 PM, Stefan Bellon wrote: > Is subversion equally frowned upon? If not, what about: > > fossil diff --from PREV --to HEAD myfile.c > That would be too easy ;). Yes, it covers the most common case, but it would be much more interesting to be able to say "go back N versions", as such a feature implies the "prev" behaviour as well. > The question is whether PREV and HEAD (and perhaps other symbolic names) > should be bound to one branch (i.e. the one of the current checked out > tree) or whether it should be global and possibly cross branches. > That's a good point. My gut feeling is that the currently-checked-out branch "should" be used, defaulting to trunk if there is no checkout (that said, i think we once had a thread where the user had removed or renamed his trunk branch to something else). Or maybe a syntax like branch_name:symbolic_name (which of course only works as long as nobody uses ':' in their branch names). -- - stephan beal http://wanderinghorse.net/home/stephan/ http://gplus.to/sgbeal ___ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Couple of newbie questions on artifacts and comparing
On Tue, 08 Jan, Stephan Beal wrote:
> On Tue, Jan 8, 2013 at 2:02 PM, Gilles wrote:
>
> > "fossil gdiff last beforelast myfile.c"
>
> A similar request has come up several times, and is certainly
> something we should consider (this would affect commands other than
> diff, e.g. checkout could also use this). i'd be up for programming
> it if we (==the list) can agree on syntax/semantics and we don't have
> to support all of the 3 million options supported by git ("go up 2
> revs, sideways 14, then down another 2"). (That said, i'm probably
> exaggerating - i only briefly read about this feature in git.)
Is subversion equally frowned upon? If not, what about:
fossil diff --from PREV --to HEAD myfile.c
In any case I think the --from and --to should be required and only
symbolic names for the revision hashes should be introduced.
The question is whether PREV and HEAD (and perhaps other symbolic names)
should be bound to one branch (i.e. the one of the current checked out
tree) or whether it should be global and possibly cross branches.
Greetings,
Stefan
--
Stefan Bellon
___
fossil-users mailing list
[email protected]
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Couple of newbie questions on artifacts and comparing
On Tue, Jan 8, 2013 at 2:02 PM, Gilles wrote:
> "fossil gdiff last beforelast myfile.c"
>
A similar request has come up several times, and is certainly something we
should consider (this would affect commands other than diff, e.g. checkout
could also use this). i'd be up for programming it if we (==the list) can
agree on syntax/semantics and we don't have to support all of the 3 million
options supported by git ("go up 2 revs, sideways 14, then down another
2"). (That said, i'm probably exaggerating - i only briefly read about this
feature in git.)
--
- stephan beal
http://wanderinghorse.net/home/stephan/
http://gplus.to/sgbeal
___
fossil-users mailing list
[email protected]
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Couple of newbie questions on artifacts and comparing
On Tue, 8 Jan 2013 13:43:22 +0100, Stefan Bellon wrote: >> Next I tried the first artifact ID: >> fossil gdiff --from 53ff5087e3 --to ad6f12df62 >> >> This works, but it seems to go through all the files that were part of >> each commit. > >fossil diff --from 53ff5087e3 --to ad6f12df62 myfile.c Thanks. I was under the wrong impression that the artifact ID was enough to identify a specific revision of a specific file. I would suggest adding the shortcuts to (g)diff: "fossil gdiff last beforelast myfile.c" Thank you. ___ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Couple of newbie questions on artifacts and comparing
On Tue, 8 Jan 2013 13:38:06 +0100, Stephan Beal wrote: >> 2012-11-25 [ad6f12df62] New version (user: Joe, >>artifact: [dc79071933], branch: trunk) >> 2012-11-25 [53ff5087e3] Original file (user: Joe, >>artifact: [54dff7fa29], branch: trunk) >> >Correct. It's not immediately obvious, but the first number is the commit >UUID and the second is the file artifact itself: > >zOut = sqlite3_mprintf( > "[%.10s] %s (user: %s, artifact: [%.10s], branch: %s)", > zCiUuid, zCom, zUser, zFileUuid, zBr); Thanks. If we must use the commit UUID when comparing different revisions of a file, In which case do we need the file artifact ID then? ___ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Couple of newbie questions on artifacts and comparing
On Tue, 08 Jan, Gilles wrote: > Next I tried the first artifact ID: > fossil gdiff --from 53ff5087e3 --to ad6f12df62 > > This works, but it seems to go through all the files that were part of > each commit. fossil diff --from 53ff5087e3 --to ad6f12df62 myfile.c Greetings, Stefan -- Stefan Bellon ___ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Couple of newbie questions on artifacts and comparing
On Tue, Jan 8, 2013 at 1:02 PM, Gilles wrote: > 2012-11-25 [ad6f12df62] New version (user: Joe, >artifact: [dc79071933], branch: trunk) > 2012-11-25 [53ff5087e3] Original file (user: Joe, >artifact: [54dff7fa29], branch: trunk) > ? > > In the output above, are "53ff5087e3" and "ad6f12df62" the artifact ID > of the whole commit this file was part of, while "54dff7fa29" and > "dc79071933" are the artifact ID of the two revisions of this > particular file? > Correct. It's not immediately obvious, but the first number is the commit UUID and the second is the file artifact itself: zOut = sqlite3_mprintf( "[%.10s] %s (user: %s, artifact: [%.10s], branch: %s)", zCiUuid, zCom, zUser, zFileUuid, zBr); > How can I get gdiff (WinMerge) to compare two revisions, usually the last and before-last revisions? > If i'm interpreting the 'diff' help properly... try: fossil gdiff --from 53ff5087e3 --to ad6f12df62 FILE1 ... FILEN :-? -- - stephan beal http://wanderinghorse.net/home/stephan/ http://gplus.to/sgbeal ___ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
