refactor: improve compaction logic

[tri-adam]

Improve the behaviour of image compaction when adding/deleting data objects.

Improve compaction when OptDeleteCompact is supplied to DeleteObject. Do not fail when compaction is requested on a data object that is not at the end of the data section. When compaction is requested, always calculate the end of the data section based on the in-use descriptors.

Previously, AddObject would simply append the data object to the end of the data section. If one or more object(s) had previously been deleted from the image, this could result in wasted space. Now, we trim the data section based on in-use objects before adding the new object.

Add/update tests to reflect new behaviour.

Closes #193

[codecov]

Codecov Report

Attention: Patch coverage is 83.33333% with 3 lines in your changes missing coverage. Please review.

Project coverage is 62.64%. Comparing base (e3aa617) to head (e8dad67).

Files	Patch %	Lines
pkg/sif/delete.go	57.14%	2 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #373      +/-   ##
==========================================
+ Coverage   62.59%   62.64%   +0.05%     
==========================================
  Files          37       37              
  Lines        3240     3234       -6     
==========================================
- Hits         2028     2026       -2     
+ Misses       1046     1043       -3     
+ Partials      166      165       -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[dtrudg]

This looks sensible to me, but I'm missing something....

What is now causing these to be 3 bytes shorter? I'm clearly missing the cause...

[tri-adam]

What is now causing these to be 3 bytes shorter? I'm clearly missing the cause...

That's a result of this line:

https://github.com/tri-adam/sif/blob/e8dad67f88b7af7d7d3d4933d1c301428c17a319/pkg/sif/create.go#L97

Those two test cases delete the last object in the SIF without supplying the compact option, and then add a new object. Previously, the code would append after the end of the data section (which still included the deleted object.) The new code calculates where the end of the valid data actually is, and appends there.

[dtrudg]

What is now causing these to be 3 bytes shorter? I'm clearly missing the cause...

That's a result of this line:

https://github.com/tri-adam/sif/blob/e8dad67f88b7af7d7d3d4933d1c301428c17a319/pkg/sif/create.go#L97

Those two test cases delete the last object in the SIF without supplying the compact option, and then add a new object. Previously, the code would append after the end of the data section (which still included the deleted object.) The new code calculates where the end of the valid data actually is, and appends there.

Thanks... apologies I missed that!