Re: Info reader fails to find cross-references to anchors

[Gavin Smith]

On 18 July 2015 at 20:23, Eli Zaretskii <address@hidden> wrote:
> Thanks.  I guess what I don't quite understand is the duplication of
> the node info between TAG and its constituent cache, which is a NODE.
>
> Also, if the cache's purpose is to cache information, why does
> find_node_from_tag erases that information?

Say we open an Info file with CR-LF line endings where the CR bytes
are not counted in the tag table. The first node we open we manage to
find using the small slack (about 4 bytes), acting as if the CR bytes
are counted. But later we try to open another node and can't find it.
What we do then is strip out all the CR's before LF's so the tag table
will be correct. But the locations of any nodes we found before will
be relative to the original contents before the stripping, so we
should forget about them and look for them again. That's why I threw
away all the information about the node. Note that this happens at
most once per opened Info file.

It might indeed be possible to keep some of the information in the tag
table (e.g. Next and Prev won't change because of end-of-line
conversion), and also there may be duplicated info (e.g. looking at
nodes.h I can't remember what the 'flags' field of TAG is for).