[pbs-devel] [RFC pxar 1/20] fix #3174: encoder: impl fn new for LinkOffset

[Christian Ebner]

Agreed, I will introduce a separate type for appendix reference offsets as well
as for other offsets, if needed, leaving the LinkOffset as is.

> On 28.09.2023 08:49 CEST Wolfgang Bumiller <w.bumiller at proxmox.com> wrote:
> 
>  
> On Wed, Sep 27, 2023 at 02:26:03PM +0200, Christian Ebner wrote:
> > > On 27.09.2023 14:08 CEST Wolfgang Bumiller <w.bumiller at proxmox.com> wrote:
> > > 
> > >  
> > > On Fri, Sep 22, 2023 at 09:16:02AM +0200, Christian Ebner wrote:
> > > > Allows to generate a new LinkOffset for storing the offset of regular
> > > > files in the backup catalog, based on the provided archiver position.
> > > > 
> > > > Signed-off-by: Christian Ebner <c.ebner at proxmox.com>
> > > > ---
> > > >  src/encoder/mod.rs | 6 ++++++
> > > >  1 file changed, 6 insertions(+)
> > > > 
> > > > diff --git a/src/encoder/mod.rs b/src/encoder/mod.rs
> > > > index 0d342ec..710ed98 100644
> > > > --- a/src/encoder/mod.rs
> > > > +++ b/src/encoder/mod.rs
> > > > @@ -31,6 +31,12 @@ pub use sync::Encoder;
> > > >  pub struct LinkOffset(u64);
> > > >  
> > > >  impl LinkOffset {
> > > > +    /// Create a new link from the raw byte offset.
> > > > +    #[inline]
> > > > +    pub fn new(raw: u64) -> Self {
> > > 
> > > Not very happy about that. It was meant to be an opaque type that you
> > > can't just create easily.
> > > But oh well... if we need it...
> > > Better than using u64 directly.
> > 
> > Might it be worth to create dedicated types or enums for the different offsets
> > needed? I do need some way to hande the different offsets and this just seemed
> > better than some raw u64, in order to have at least some type checking.
> 
> For different purposes I would *prefer* different types, if
> possible.
> Since your "new" offsets come from a new method (`app_appendix`), it
> should IMO be possible.
> That way it's clear that you shouldn't use these offsets for eg.
> `add_hardlink` and you bind the value to a purpose via a type.