Remove some assertions that are not necessarily true (fixes #514)

Operations that add the same object to multiple places in the pages
tree are throwing exceptions and then later causing assertion
failures. The assert calls shouldn't be there.

1 files changed, 27 insertions, 0 deletions

diff --git a/TODO b/TODO
index 3b86e3db..2d03e0be 100644
--- a/TODO
+++ b/TODO
@@ -1,6 +1,12 @@
 Document-level work
 ===================
 
+* Implement the tree helper described in the ABI section and provide
+  the getPagesTree method so QPDF::getAllPages can be deprecated in
+  10.4 and people have time to transition before qpdf 11. Maybe I
+  should have a qpdf 11 branch so I can make sure I have the
+  interfaces the way I want them before releasing 10.4.
+
 * QPDFPageCopier -- object for moving pages around within files or
   between files and performing various transformations
 
@@ -182,6 +188,25 @@ Comments appear in the code prefixed by "ABI"
   before copying, though maybe we don't because it could cause
   multiple copies to be made...usually it's better to handle that
   explicitly.
+* (Also API) Create a helper class for tree structures like the pages
+  tree that have /Kids and /Count. It would be great to be able to
+  eliminate flattenPagesTree and all the page object to page caching,
+  and to get rid of getAllPages() that returns a reference to an
+  internal data structure. The tree helper object can have
+  bidirectional iterator and const_iterator, implement at and
+  operator[], and have reasonable insertion and deletion semantics.
+  Then a new getAllPages, perhaps getPagesTree, can return that and
+  make it easy to change existing code even if it assumes the result
+  of getAllPages() is mutating automatically as the pages tree is
+  manipulated. This will free us up to finally get rid of the
+  troublesome pages cache and its exposure by getAllPages. See also
+  note below with "balance the pages tree". Maybe we can get rid of
+  pageobj_to_pages_pos as well or convert it into something that is
+  merely advisory and not so heavily trusted. We should either make
+  the code not care if the same object appears more than once in the
+  pages tree or detect it and shallow copy, though the latter would
+  cause a significant performance penalty when manipulating pages if
+  we get rid of the cache.
 
 Page splitting/merging
 ======================
@@ -334,6 +359,8 @@ directory or that are otherwise not publicly accessible. This includes
 things sent to me by email that are specifically not public. Even so,
 I find it useful to make reference to them in this list.
 
+ * Get rid of remaining assert() calls from non-test code.
+
  * Consider updating the fuzzer with code that exercises
    copyAnnotations, file attachments, and name and number trees. Check
    fuzzer coverage.