X-Git-Url: http://developer.intra2net.com/git/?p=libt2n;a=blobdiff_plain;f=codegen%2FTODO;h=eb7631c3c898148453a6e61a26b31ab11883f7ac;hp=601a5a0d4eb7281fce81d60ca4c57aeb097f7e55;hb=a63e08b83794273da3840e0b8bf15bf0085fe3c4;hpb=a930cc995e41002986df3d6c9b722eea11fd6627

diff --git a/codegen/TODO b/codegen/TODO
index 601a5a0..eb7631c 100644
--- a/codegen/TODO
+++ b/codegen/TODO
@@ -1,30 +1,11 @@
-- remove support for LIBT2N_EXPORT_GROUP(group) (this will simplify the generator a lot)
 - add option similar to gccs -MD (at the moment it is a fixed set of files generated and
   they are handled in the makefile snippet [codegen.make])
-
-open questions:
-- should projects using the codegen depend on installed version of ... or ship their own version?
-- get rid of codegen-stubhead.hxx or include a "copy" in each project?
-  first we said yes but now i say no because the lib depends on the libt2n headers anyway
-- makefile snippet must work for builds outside of libt2n
-  => some variables must be set by configure
-  the variables which must be set:
-	LIBT2N_CODEGEN="\$(top_builddir)/codegen/libt2n-codegen"
-	LIBT2N_CLIENT_PCTEMPLATE="\$(top_srcdir)/codegen/clientlib.pc.in"
-  => we store the variables in the .pc file of libt2n
 - it would really be much nicer if the client lib includes would not depend upon boost serialization
   i thought a solution would be to provide this optionally by wrapping the corresponding includes in a #ifdef
   but this does not work since command.hxx is included which depends on boost serialization headers anyway
   => we do not provide this for now
 - naming scheme?!
   perhaps generated group class should not be prefixed by cmd_group_
-
-docu
-- for the docu we need a minimal example
-- for automatic testing we need a complex example
-
-the example should show:
-- how to mark procedures
-- how to write configure.ac and Makefile.am
-- that we create a library for the client
-- how to use that library
+- would be nice to reduce amount of changed files
+  (codegen output is often the same as the existing files - something similar to ccache
+  for codegen - but not overwriting output files if they are the same)