On Dec 12, 2007, at 3:09 PM, Jeff Johnson wrote:

> This patch likely needs some explanation too.
>
> rpm-5.0 looks up tag names using a binary, not a linear, search. That
> has to be a win, and its on mainline header tag access paths. However,
> the effect may be kinda hard to measure objectively, most accesses
> just need one or two tags.
>

One last attempt to measure the effect (still not the right benchmark,
I was expecting a bigger win):

[jbj@wellfleet wdj]$ rpm -qia --stats --nodigest --nosignature -- 
tagbsearch > /dev/null
    total:               1      0.000000 MB      2.143929 secs
    dbget:            1285     34.413912 MB      0.146196 secs
    hdrload:          1283     34.413908 MB      0.168050 secs
    hdrget:          23512      0.000000 MB      0.032874 secs
[jbj@wellfleet wdj]$ rpm -qia --stats --nodigest --nosignature -- 
notagbsearch > /dev/null
    total:               1      0.000000 MB      2.123044 secs
    dbget:            1285     34.413912 MB      0.146461 secs
    hdrload:          1283     34.413908 MB      0.166587 secs
    hdrget:          23512      0.000000 MB      0.032235 secs
[jbj@wellfleet wdj]$ rpm -qia --stats --nodigest --nosignature -- 
tagbsearch > /dev/null
    total:               1      0.000000 MB      2.149398 secs
    dbget:            1285     34.413912 MB      0.145321 secs
    hdrload:          1283     34.413908 MB      0.166040 secs
    hdrget:          23512      0.000000 MB      0.031639 secs
[jbj@wellfleet wdj]$ rpm -qia --stats --nodigest --nosignature -- 
notagbsearch > /dev/null
    total:               1      0.000000 MB      2.169871 secs
    dbget:            1285     34.413912 MB      0.144984 secs
    hdrload:          1283     34.413908 MB      0.166574 secs
    hdrget:          23512      0.000000 MB      0.032085 secs

Back to fixing instead of goofing around ...

73 de Jeff