From: pgstef (@pgstef) <noreply+pgstef@github.com>
To: pgbackrest/pgbackrest <noreply+pgbackrest-pgbackrest@github.com>
Subject: Re: [pgbackrest/pgbackrest] PR #2928 (approved): Zero block incremental info for truncated files.
Date: Thu, 19 Nov 2026 09:30:27 +0000
Message-ID: <gh-pgbackrest-pgbackrest-2928-r5972434672@github.com> (raw)
In-Reply-To: <gh-pgbackrest-pgbackrest-2928@github.com>
References: <gh-pgbackrest-pgbackrest-2928@github.com>
The PR looks good to me! 👍
I think I noticed one small issue during the review, though. It's not introduced by this PR, but I thought it might be worth mentioning since we're touching block incremental metadata.
If I understand correctly, the `manifest` command still checks `blockIncrSize != 0` when deciding whether to render block incremental information, rather than checking `blockIncrMapSize != 0`.
This would mean that manifests created by earlier versions, where a file could have a non-zero `blockIncrSize` but no block map, could still cause problems. Without `--filter`, the command would display `map size 0B`, while using `--filter` on such a file would attempt to read the missing `.pgbi` file and fail with a `FileMissingError`.
I think we could simply check `blockIncrMapSize != 0` instead in `cmdManifestFileRender()`, for both JSON and text output. That would also make the rendering consistent with the actual presence of a block map.
It's a pre-existing issue, but I thought it was worth pointing out.