Re: [Libcdio-devel] About old branches of mine

[Rocky Bernstein]

Over the weekend I removed my stale branches from libcdio's git.

On Fri, Mar 31, 2023 at 5:25 PM Thomas Schmitt <scdbackup@gmx.net> wrote:

> Hi,
>
> Rocky Bernstein
> >     test/data/multi_extent_8k.iso        | Bin 524288 -> 122880 bytes
> > I think both versions should be there. We should be testing not just how
> > things are created currently, but also how they may have been created in
> > the past.
>
> I realize that i forgot to take into account your mail of
> Mon, 27 Mar 2023 09:11:32 -0400.
>
> But actually i wanted to ask back why you want to keep an ISO which was
> just created a bit wasteful for its purpose.
>
> It will never end up on a CD that gets written with write type TAO. Thus
> it needs no 300 KiB of padding.
> It will never be enlarged by multi-session. Thus a separate superblock
> for the first session is not needed.
>

The goal in testing is to try all sorts of combinations that can occur.

if xoriso or some other burner program in the past created ISO 9660 images
in a particular way, then testing
should be verifying that it can read those kinds of images.

One can try to reason or argue that the exactly the same code path is taken
in libiso9660, but it is possible to make a mistake,
even if it is indeed the situation here. Better then is just to try both
kinds of images. It is not like we are already testing a large number of
images already.



> The padding compensates for mislead reading ahead by Linux when the
> drive firmware tells questionable read capacity which is invited by a bug
> in the SCSI specs.
>
> The separate superblock would survive the writing of a next session and
> stay mountable. Also it would be the starting point of the search for the
> chain of session superblocks. xorriso command -toc can then tell a session
> history like
>   ISO session  :   1 ,        32 ,   2024199s , HOME_2022_07_07_155642
>   ISO session  :   2 ,   2024256 ,     26157s , HOME_2022_07_08_155953
>   ISO session  :   3 ,   2050432 ,     32162s , HOME_2022_07_09_151801
>   ...
>   ISO session  : 267 ,  11491200 ,     52690s , HOME_2023_03_30_193636
>   ISO session  : 268 ,  11543904 ,     44883s , HOME_2023_03_31_154321
>   Media summary: 268 sessions, 11584524 data blocks, 22.1g data,  976m free
>
> Linux or libcdio are supposed to use the first superblock at LBA 0 which
> always points to the newest session. So the old test ISO and the new one
> are equivalent as far as libcdio is concerned.
> If we keep the old ISO then there is no need to have the new one. It would
> only be more waste of space.
>
>
> Have a nice day :)
>
> Thomas
>
>
>