Colin Rosenthal

Follow-ups committed after a little bit of back-and-forth.

Follow-ups committed after a little bit of back-and-forth.

I could just have checked only that it is not a directory (and not bothered with the symlink check). That would have skipped some possible error cases - e.g. if you accidentally specify a file inst...

I could just have checked only that it is not a directory (and not bothered with the symlink check). That would have skipped some possible error cases - e.g. if you accidentally specify a file instead of a directory then it should throw an exception.

Good catch. The Annotation says this line is from Kåre Fiedler Christiansen from 2009!

Good catch. The Annotation says this line is from Kåre Fiedler Christiansen from 2009!

I agree that the if/else construction is more transparent. I don't agree about removing the return statement. If the file is a symbolic link and not a directory (ie unresolved) then we just want t...

I agree that the if/else construction is more transparent.

I don't agree about removing the return statement. If the file is a symbolic link and not a directory (ie unresolved) then we just want this method to return without doing anything. If it's not a symbolic link then we carry on to the subsequent checks (that it exists, is a directory, is writable).

(The code is tested and works as is.)

Changes to support dkupb-109
Changes to support dkupb-109
To support https://sbprojects.statsbiblioteket.dk/jira/projects/DKUPB/issues/DKUPB-109?filter=allopenissues need to be able to start a NetarchiveSuite instance where the bitstorage directories are ...

To support https://sbprojects.statsbiblioteket.dk/jira/projects/DKUPB/issues/DKUPB-109?filter=allopenissues need to be able to start a NetarchiveSuite instance where the bitstorage directories are dangling links (which bitmag undangles to directories during storage). But the current implementation is very defensive and always checks that the configured storage directories exist and are directories.

Note that our ultimate goal is to drop the NetarchiveSuite Bitarchive implementation completely so this code is not expected to go into production.

This is a very good point. Review of Bitrepository class and any related code is now added to our new Grevlings tech-deficit backlog for prioritisation in the next sprint.

This is a very good point. Review of Bitrepository class and any related code is now added to our new Grevlings tech-deficit backlog for prioritisation in the next sprint.

Accepted and resolved.

Accepted and resolved.

so add a trim() as well

so add a trim() as well

NAS-2848 Default Collection for NAS/bitmag
NAS-2848 Default Collection for NAS/bitmag
Minor changes only required (translation)

Minor changes only required (translation)

Correct translation

Correct translation

NAS-2792: Add purple bullet png
NAS-2792: Add purple bullet png
javadoc

javadoc

remove the comments

remove the comments

fix javadoc

fix javadoc

Actually it doesn't matter. The getHost() method will trim the scheme away anyway. However without the scheme, the "getInstance()" method will die. It might be better to add "dummyscheme://" instea...

Actually it doesn't matter. The getHost() method will trim the scheme away anyway. However without the scheme, the "getInstance()" method will die. It might be better to add "dummyscheme://" instead, to show that the scheme doesn't matter.

But not today.

But not today.

Fix this description

Fix this description

finish javadoc

finish javadoc

remove comment.

remove comment.

Using the new StringJoiner utility makes it easier to build a String separated by newlines. Consider the possibility.

Using the new StringJoiner utility makes it easier to build a String separated by newlines. Consider the possibility.

remove these commented lines.

remove these commented lines.

We should try to make the new placeholders more consistent in their naming. I think they should all start with "UMBRA_" so they are easy to find in the cxml.

We should try to make the new placeholders more consistent in their naming. I think they should all start with "UMBRA_" so they are easy to find in the cxml.

javadoc

javadoc