Skip to content

Make trace files lexicographically ordered #1825

@mpilman

Description

@mpilman

Currently trace-files have the following format:

trace.[IP-Address].[TIMESTAMP].[RND-String].[SEQ-NUMBER].xml

This means that they are not lexicographically ordered. So if you grep within a log directory (or use globs on bash or simply ls), the 10th file will show up before the 2nd file.

I would like to change this format so that all files that a process writes are in order. One very simple possible way of doing this would be to make the sequence number fixed width (so 1 would become 0000000001). I know that this might make the whole thing look a bit ugly.

My preferred solution would be the following: Instead of using the above trace name format we could use the following trace name format:

trace.[IP-Address].[TIMESTAMP].[RND-String].[WIDTH-OF-SEQ-NUMBER - 1].[SEQ-NUMBER].xml

I think it is safe to assume that we will never have a process that creates 9999999999 trace files. The trace files will look almost the same and they will be lexicographically ordered.

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Relationships

None yet

Development

No branches or pull requests

Issue actions