- From: Norman Walsh <ndw@nwalsh.com>
- Date: Fri, 25 Jan 2008 10:35:57 -0500
- To: public-xml-processing-model-comments@w3.org
- Message-ID: <m2ejc5nciq.fsf@nwalsh.com>
/ Vasil Rangelov <boen.robot@gmail.com> was heard to say: |>I think we've addressed this issue. | | OK... how? The original issue was about p:exec, which we added. :-) |>From what I have gathered from the XProc minutes from 20 Dec 2007, | environmental variables won't be supported (a bummer, but I can live with | that), and as such, they are explicitly not allowed in either the "cmd" or | the "args" options. | | In other words, an implementation running on Windows must not convert the | path "%WINDIR%\explorer.exe" to "C:\WINDOWS\explorer.exe", but instead, | accept it as an invalid path. The same goes for environment variables in | "args". The implementations must pass "%WINDIR%" as the string "%WINDIR%", | hoping the windows program itself will resolve it, not resolve it itself to | "C:\WINDOWS" and then pass that down to the program. Right. | While I don't mind this behavior, it should still be explicitly stated in | the spec, and I don't see it in the latest one (10 January 2008) yet. Ok, I just made that explicit. | Offtopic:// | I must say I'm all for the top level change. p:declare-step always felt | useless up until now. :-) Be seeing you, norm -- Norman Walsh <ndw@nwalsh.com> | If man were never to fade away like the http://nwalsh.com/ | dews of Adashino, never to vanish like | the smoke over Toribeyama, but lingered | on forever in the world, how things | would lose their power to move us! The | most precious thing in life is its | uncertainty.--Yoshida Kenko
Received on Friday, 25 January 2008 15:36:12 UTC