[rb-general] SOURCE_PREFIX_MAP format specification proposals

[Daniel Kahn Gillmor]

On Sat 2017-01-14 05:45:00 -0500, Ximin Luo wrote:
> I agree that the principles expressed are very important for a
> communications protocol or encoding system, but this is not what
> SOURCE_PREFIX_MAP is. The only way it will be transmitted, is via
> .buildinfo files, which already quote all envvar values (see
> deb-buildinfo(1)).

It may also be transmitted in documentation, articles in engineering
magazines, development-focused wikis, etc.  Ignoring these transmission
vectors is possible, but seems likely to limit understanding and
adoption.

> PATH does not escape colons, it just disallows paths-with-colons in PATH.

I like the simplicity of this approach.  Is there any reason we need
to support paths-with-colons?  what about LD_LIBRARY_PATH?  If we can
just accept this already-shared limitation, it makes implementation and
use *much* simpler because there is no escaping to worry about at all.

> I had another idea last night which is to use urlencoding, so it would look like this:
>
> SOURCE_PREFIX_MAP=/a/b%3Dyyy=ERROR&/a=lol&/b=foo&/a/b%3Dyyy=secret

Bundling a URL-decoder in a toolchain that currently doesn't need to
have such a thing sounds like something that would make me sad, but
maybe i'm too conservative.

        --dkg
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.reproducible-builds.org/pipermail/rb-general/attachments/20170114/c161631c/attachment.sig>