Re: [PATCH v3 02/15] xdiff: introduce rust

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Phillip Wood <phillip.wood123@xxxxxxxxx> writes:

>     This release introduces an optional dependency on rust that is
>     enabled by default. Platforms without a rust compiler can continue
>     to build git by passing NO_RUST=1. In six months time we plan to
>     make rust mandatory for building git. From that point git 2.x.y (the
>     last version that can be built without rust) will continue to
>     receive security updates for three years.
>
> To me the important elements are:
>
> 1) There is a short period where rust is optional. This allows
>    (i) Distributors on platforms without a rust compiler time to notify
>        their users that in the future they will only be able to offer
>        security updates.
>   (ii) Distributors on platforms with a rust compiler time to adjust
>        their build procedures to include rust.
>  (iii) The git project time to gain experience of using rust and writing
>        the necessary bindings while building with it is optional.

Good.  I am not sure "short" should be an important element, but
having a known and agreed-upon deadline helps.

> 2) Rust is enabled by default so platforms without a rust compiler are
>    made aware of the problem but have an easy way to continue to build
>    git while rust is optional.

Obviously there is nothing to disagree with here, as it is the
definition of the word "optional" ;-).

> 3) There is a period of a small number of years where we continue to
>    provide security updates for a version of git that can be built
>    without rust. This is intended to  allow a realistic time for
>    distributors on platforms without a rust compiler to port one or make
>    other arrangements for providing future security updates without
>    placing an undue burden on the project to provide security updates
>    for niche platforms indefinitely.

I am not willing to see such a support for multiple years, though.
If the first item is 6 months, this backporting stale releases
should be on the same order of timeperiod.

If it were "3 years of optional period, 18 months of backporting
security updates", I would find it more realistic.  It would give
those platform maintainers enough time to robby, fundraise, or
otherwise campaign to bring Rust on their system.  I personally find
that 6 months is way too short (if we are _only_ looking for an
excuse to say "we have given them ample time to react, and now it is
their problem", 6 months may be good enough, though).





[Index of Archives]     [Linux Kernel Development]     [Gcc Help]     [IETF Annouce]     [DCCP]     [Netdev]     [Networking]     [Security]     [V4L]     [Bugtraq]     [Yosemite]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux SCSI]     [Fedora Users]

  Powered by Linux