Re: question about anchor database

[Shigio Yamaguchi]

> In creation of an anchor database, it has taken time rather than creation of 
tag files.
> I think that creation of an anchor database takes time too much.

You are right.

> And I want to know what is a bottleneck.
> The form of tag files influences the performance of global -nx.
> As the last mail showed, the difference was seldom seen.

Htags use anchor database for picking up all tags which belong to a file.
But db(3) library has only one index, we cannot read above information
from tag files efficiently. (Yes, we need to read record by file name
as a key, not by tag name.)
In compact format, we can do the job by reading less records.

> Which does time require, the writing to an anchor database, and execution of 
PERL?

Do you mean 'in memory' by the word 'PERL'?
If we have enough memory then we had better do all the job in memory.
But in poor environment or in the system which has many users, it may brings
a breakdown. So, huge data should be written to disk, I think.

I think the best method is to use new database library which afford multi-key.
But it cut off upper compatibility. :(
--
Shigio Yamaguchi <address@hidden> - Tama Communications Corporation
Spare mail address: <address@hidden>
PGP fingerprint: D1CB 0B89 B346 4AB6 5663  C4B6 3CA5 BBB3 57BE DDA3