Re: Info reader fails to follow xrefs to anchors

[Gavin Smith]

On 13 July 2015 at 21:21, Ludovic Courtès <address@hidden> wrote:
> File attached.  As an example: Go to “Substitutes” and try to follow the
> anchors ‘daemon-substitute-urls’ and ‘client-substitute-urls’ (see
> <http://git.savannah.gnu.org/cgit/guix.git/tree/doc/guix.texi#n1439>.)

Here's the fix. I expect it could be used with little modification on
the 6.0 release. What I am going to commit to the development version
will be slightly different.

Index: info/info-utils.c
===================================================================
--- info/info-utils.c   (revision 6428)
+++ info/info-utils.c   (working copy)
@@ -1611,8 +1611,9 @@

    If FB is non-null, it is the file containing the node, and TAG_PTR is an
    offset into FB->tags.  If the node contents are rewritten, adjust anchors
-   that occur in the node. */
-
+   that occur in the node and store adjusted value as TAG->nodestart_adjusted,
+   otherwise simply copy TAG->nodestart to TAG->nodestart_adjusted for each
+   anchor in the node. */
 void
 scan_node_contents (NODE *node, FILE_BUFFER *fb, TAG **tag_ptr)
 {
@@ -1749,6 +1750,16 @@
          null, that is, the length. */
       node->nodelen = text_buffer_off (&output_buf) - 1;
     }
+  else if (fb && tag_ptr)
+    {
+      /* Set nodestart_adjusted for all of the anchors in this node. */
+      tag_ptr++;
+      while (*tag_ptr && (*tag_ptr)->nodelen == 0)
+        {
+          (*tag_ptr)->nodestart_adjusted = (*tag_ptr)->nodestart;
+          tag_ptr++;
+        }
+    }
 }