Benchmark: Git 1.6.3.3, Hg 1.3.1, Bzr 1.17

[David Ingamells]

For completeness have stacked branches been tried where you have said 
"clone"????
Performance /ought/ to be similar to lightweight checkouts.

David.

Jari Aalto wrote:
> Robert Collins <robertc at robertcollins.net> writes:
>
>   
>> Jari, can you try this for the bzr setup:
>>
>> bzr init-repo --2a --no-trees
>> bzr init trunk
>> bzr checkout --lightweight trunk working
>> # do stuff in working
>> bzr branch ../trunk ../newbranch; bzr switch newbranch # this is 'clone'
>> # do more stuff in working
>>     
>
> Tested and added to the test script. Indeed branching is quite fast (See *).
>
> But that seems like very complex setup for regular users. Compared to
> straighforward "hg branch NAME" and "git branch NAME".
>
> I've updated the data[1] in this respect:
>
>     Kernel 2.6.30 import
>                                                                     Shard repo
>                                                                     + --lightweight
>                     Git 1.6.3.3     Hg 1.3.1        Bzr 1.17 --2a
>     --------------------------------------------------------------+------
>     init            0,03            0,05            0,90     0,90 |  1,50
> 1   add             26,00           1,80            5,50     5,50 |  6,20
>     commit          0,60            51,00           55,00    53,00| 62,00
>
>     ...continue with 2.6.29 patch, 7552 files
>
>                     Git 1.6.3.3     Hg 1.3.1        Bzr 1.17 --2a         Bzr vs. Hg
>                                                                           (% bigger)[2]
>     --------------------------------------------------------------+------ -------------
> 2   add             6,30            0,50            3,00    3,5   |  3,50 600
>     commit          0,50            28,00           37,00   24,00 | 25,00 -11
>     log             0,01            0,13            0,17    0,19  |  0,20  54
> *   branch          0,01            0,10            -       -     |  0,50 400
>     clone           49,00           40,00           71,00   71,00 | 71,00  78
>
> Boths "add" commands (1, 2) in Bzr are less than Hg's.
>
> Concerning the "2.6.29 patch" data, by using the latest
> shared-repo/lightweight/switch, Bzr is in average:
>
>     - 224% slower (all rows)
>     - 140% slower than Hg (if we ignore 'add' row value 600%)
>     - 40% slower (if we ignore 'add' and 'branch' rows)
>
> Presumed I calculated the values correctly in editgrid. Please spot
> errors if you see any.
>
> Jari
>
> [1] http://www.editgrid.com/user/jaalto/vc-test
> [2] Formula for "How much Y is bigger than X": (Y - X) / X * 100
>     X = Hg
>     Y = Bzr
>
>
>