- From: Norman Walsh <ndw@nwalsh.com>
- Date: Thu, 09 Apr 2009 14:11:58 -0400
- To: public-xml-processing-model-comments@w3.org
- Message-ID: <m2tz4xu26p.fsf@nwalsh.com>
Toman_Vojtech@emc.com writes: > I have a question about p:exec. Section 7.2.1 says: > > "If the command or cwd options contain any '/' or '\' characters, they > will be replaced with the platform-specific path separator character. If > the fix-slashes option is true, this fixup will be applied to args as > well." > > What if I get totally crazy and put the executable in a directory that > has a backshlash in the name in a *nix like OS? Replacing of path > separators may break things in that case. > > -- > > Also, how does the 'fix-slashes' option work with this: > > <p:exec args="snake='\/\/\/\O<'">...</p:exec> > > Does this 'fix' the slashes in the escaped argument value? At the 9 Apr 2009 telcon, the WG decided to pursue a quite different approach to this problem. On the one hand, that's kind of risky in CR, on the other, we've already convinced ourselves that we could *add a new port* to the p:exec step in CR, so I don't see how this could be described as more significant. Instead of a boolean option, we're going to change the fix-slashes option so that it is a single character. If specified, then any *and all* occurrences of that character in both the command and the args will be replaced by the platform-specific slash. This means that processing of the command and args will no longer be independent. But it also means you don't have nasty problems like the one Vojtech identifies above. The editor will post a concrete proposal as soon as possible. Be seeing you, norm -- Norman Walsh <ndw@nwalsh.com> | Wink at small faults; for thou has http://nwalsh.com/ | great ones.--Thomas Fuller (II)
Received on Thursday, 9 April 2009 18:12:40 UTC