Re: coding an Application in perl

On 02/24/16 16:41, Jean-Louis Martineau wrote:
> On 23/02/16 03:48 PM, Chapman Flack wrote:
>> Another building-block application would be something like
>> amOpaqueTree ... think of a directory tree where the files
>> and names are all under some software control and not
>> individually meaningful to you. A database data directory,
>> subversion or git repository, etc. You are not usually looking
>> for just file 1865a73f9 out of this directory, you just want
>> to recover the whole tree as it was at the time of backup.

> This could be a property of amgtar to tell it to not generate full indexing.

>> Compact incremental backups of such a tree can be made by
>> the well-known rsync tool, using its --only-write-batch

That's the key, it depends on the nature of what you're backing up,
and your objectives. If it is a system that uses an opaque directory
tree of a dozen 2 GB files, and in half of those files, *one row* has
been changed, amgtar is probably going to produce a dozen GB
backup. rsync --only-write-batch will produce about six blocks.

The tradeoff is, amgtar saves much smaller state, basically just the
last-touched dates; rsync will have local state equivalent to the
size of the tree (for *one* level; usually not nearly so much again
for additional levels). It also does more computation to find the
differences. But it doesn't need any special knowledge about what's
in the files.

Generally, even better results would be obtained by a program that
intimately knew the file format, but that would have to be totally
custom, where something based on rsync batches can be *almost* as
good, and general enough to be a generic amanda application.

It won't be what everyone wants, but the whole beauty of selectable
applications is to be able to choose the tradeoffs most desirable
for a given DLE....

-Chap