n Monday 23 of June 2014 21:20:28 Vladimir A. Pavlov wrote:
> How can we solve this? AFAIU you suggest something like reusing
> --ignore-failed-read for the feature in question (correct me if I'm wrong).

Previously, I was thinking about something like that ^^.

> But --ignore-failed-read is not exactly what I mean because:
> 1. it outputs a warning meaning "something unexpected happens":

It could for extraction also.

> 2. according to the sources it means ignoring any read errors, not
>    only "existance" ones,

That could behave the same also for extraction, if there are some
non-fatal errors.

> On the contrary, the suggested --ignore-missing suggests that:
> 1. missing member is expected so no warning is necessary when this
>    happens,

You can use fine-grained --warning configuration.

> 2. if a member exists but cannot be processed for some other reason,
>    an error must still be returned.

I can see ^^ that --ignore-failed-read does not fully meet your
expectations for semantic.

Pavel

> [...]


Reply via email to