bug-texinfo
[Top][All Lists]
Advanced

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

Re: Arbritary characters (like colons) in node names (proposal)


From: Gavin Smith
Subject: Re: Arbritary characters (like colons) in node names (proposal)
Date: Tue, 22 Apr 2014 14:07:48 +0100

On Tue, Apr 22, 2014 at 9:49 AM, Patrice Dumas <address@hidden> wrote:
> On Fri, Apr 18, 2014 at 09:28:02PM +0100, Gavin Smith wrote:
>> Hello,
>>
>> Node specifications:
>>
>> Use the same method as Next etc. in node lines. E.g. "*see
>> <label>:<nodename>," would become "*see <label>:<q3><nodename><q4>" if
>> <nodename> contained forbidden characters.
>
> The <label> needs to be quoted too, if it contains a  ̀:'.  Should the
> same quoting apply here?
>

Yes, good point. I don't see why the same quoting couldn't be used for
the reference label.

On Sat, Apr 19, 2014 at 8:52 PM, Eli Zaretskii <address@hidden> wrote:
>> The control characters would only be interpreted as quotes in
>> particular contexts. Any outside those contexts would be left as they
>> are.
>
> How do you know you are in that particular context?  Isn't that the
> very problem that causes the problems with colons?

Not really. Colon characters can occur outside of references without
any problems. There would't be a need to process the entire node
contents just to find these quoting strings: just as when following a
reference at the moment, it does not look at the entire node contents
and decide whether each colon is part of a reference or not.

On Sat, Apr 19, 2014 at 10:44 PM, Karl Berry <address@hidden> wrote:
> The quoting syntax used for images is intended to be extensible in the
> obvious way to any other use.  I think the node quoting should use it
> too.  The control character sequence it uses (address@hidden) was chosen to
> minimize the chance of affecting any terminal output in browsers that
> don't recognize it.  No one has reported any problems with it to my
> knowledge.  Worrying about real info files containing address@hidden is not
> necessary.
>

^? is ignored by terminals. The address@hidden sequence erases the previous
character, so files output directly to a terminal (with, e.g. "cat")
with have characters missing.

In fact, I tried it with the address@hidden quoting sequences: it gives info
4.13 a segmentation fault for some reason. See the attached files.

The current use of address@hidden sequences is limited: few files contain image
tags (one of the few I have on my system is the "goops" manual, in the
"Class hierarchy and inheritance of slots" node), and in index nodes,
which I expect users won't visit directly that much. This could be why
no one has reported problems with seeing them. address@hidden sequences can
still be useful for extensions like the image extension, where new
text is to be inserted into a node, or for other markers of
information.

The ^? quoting seems to be as backward-compatible as it could possibly
be: users of old readers could still open and move about old files,
except: (a) They might see the "^?" sequences quoting links to nodes
with forbidden characters in their names, and (b) They would not be
able to access such nodes.

Attachment: del-quoting.info
Description: Binary data

Attachment: null-backspace-quoting.info
Description: Binary data


reply via email to

[Prev in Thread] Current Thread [Next in Thread]