From f98a0dd6e82020efcfd947d519f0b1258cf56938 Mon Sep 17 00:00:00 2001 From: Andreas Baumann Date: Sat, 5 Nov 2011 11:42:17 +0100 Subject: more docu cleanup --- BUGS | 8 ++++++++ TODOS | 40 ++++++++++++++++++++++++++++++++++++++++ docs/makefiles/BUGS | 6 ------ docs/makefiles/TODOS | 22 ---------------------- docs/network/README | 14 ++++++++++++++ docs/port/README.networking | 7 ------- docs/port/TODOS | 10 ---------- docs/service/TODOS | 6 ------ 8 files changed, 62 insertions(+), 51 deletions(-) create mode 100644 BUGS delete mode 100644 docs/makefiles/BUGS delete mode 100644 docs/makefiles/TODOS delete mode 100644 docs/port/README.networking delete mode 100644 docs/port/TODOS delete mode 100644 docs/service/TODOS diff --git a/BUGS b/BUGS new file mode 100644 index 0000000..bbd4f10 --- /dev/null +++ b/BUGS @@ -0,0 +1,8 @@ +Known Bugs: + +Makefiles + +OpenBSD 4.3, more precise gmake 3.80 is buggy, ignore: +../makefiles/depend.mk:27: no file name for `-include' +The build process seems to be ok though.. + diff --git a/TODOS b/TODOS index 964925a..af3a89e 100644 --- a/TODOS +++ b/TODOS @@ -63,3 +63,43 @@ - Seen links with the actual event number in the event logger with the possibility to send it directly to a bug system (HTML form). This would be great to have! +- makefiles + - GNU make + - currently we are following the "recursive makes considered harmfull" :-) + - distclean calls clean again recurively. This causes already deleted + dependencies to be regenerated + - do shared libraries properly + - add uninstall + - modularize more + - get platform.mk and shell in sync, make sure the user can set the + variables there manually if we wants to + - split things for different compilers and make them individually + handleable + - check out other makefile systems: + - ION-3 + - openradius + - POCO + - problems: + - OpenBSD gmake: ../../makefiles/depend.mk:62: no file name for `-include' + - make -j test with missing testd, some dependencies are missing + - test target calls all target, this should not be necessary + - Windows: + - how to generate dependencies and include them in the NMAKE files? + - how to add the generation of a setup.exe or MSI? +- Porting + - the HAVE_XXX flags could produce a name clash with some autoconf'ed + software/library we will be using in production code. Rename to + WOLF_SYS_HAVE_XXX or similar + - feedback to the c99 snprintf guy, errors like: + - -1 instead of buffer size on boffer overflow + - extern errno in mutli-threaded code + - add a lot more tests + - %zd and similar pointers: make format macros like PRIdPTR which + get mapped to %ld. But what if snprintf could do a %zd and + fprintf not? +- Windows service + - option parsing on windows? + - have a generic message/po-file extractor, the perl script here is + an evil copy! + - handle pause/continue + - test long-duration startup and shutdown phases with checkpoint_hints diff --git a/docs/makefiles/BUGS b/docs/makefiles/BUGS deleted file mode 100644 index 4318ce5..0000000 --- a/docs/makefiles/BUGS +++ /dev/null @@ -1,6 +0,0 @@ -Known Bugs: - -OpenBSD 4.3, more precise gmake 3.80 is buggy, ignore: -../makefiles/depend.mk:27: no file name for `-include' -The build process seems to be ok though.. - diff --git a/docs/makefiles/TODOS b/docs/makefiles/TODOS deleted file mode 100644 index 8e64dbc..0000000 --- a/docs/makefiles/TODOS +++ /dev/null @@ -1,22 +0,0 @@ -- GNU make - - currently we are following the "recursive makes considered harmfull" :-) - - distclean calls clean again recurively. This causes already deleted - dependencies to be regenerated - - do shared libraries properly - - add uninstall - - modularize more - - get platform.mk and shell in sync, make sure the user can set the - variables there manually if we wants to - - split things for different compilers and make them individually - handleable - - check out other makefile systems: - - ION-3 - - openradius - - POCO - - problems: - - OpenBSD gmake: ../../makefiles/depend.mk:62: no file name for `-include' - - make -j test with missing testd, some dependencies are missing - - test target calls all target, this should not be necessary -- Windows: - - how to generate dependencies and include them in the NMAKE files? - - how to add the generation of a setup.exe or MSI? diff --git a/docs/network/README b/docs/network/README index a1c4ed6..cad30ac 100644 --- a/docs/network/README +++ b/docs/network/README @@ -44,3 +44,17 @@ Links: - async I/O on files, pipes doesn't work on Windows, there we need the Windows async event interface which doesn't look alike to select - select is very inefficient on Windows and also limited +- use POSIX getaddrinfo/getnameinfo as they provide a good top-level interface which + is quite portable (even on Windows) + +Good code examples: +- UNPV12e of W. Richard Stevens (Unix Network Programming), the must-read +- postgresql has a good getaddrinfo replacement for Windows and Cygwin +- some very helpful tips: www_kame_net_newsletter_19980604.txt +- use POSIX getaddrinfo/getnameinfo as they provide a good top-level interface which + is quite portable (even on Windows) + +Good code examples: +- UNPV12e of W. Richard Stevens (Unix Network Programming), the must-read +- postgresql has a good getaddrinfo replacement for Windows and Cygwin +- some very helpful tips: www_kame_net_newsletter_19980604.txt diff --git a/docs/port/README.networking b/docs/port/README.networking deleted file mode 100644 index c749209..0000000 --- a/docs/port/README.networking +++ /dev/null @@ -1,7 +0,0 @@ -- use POSIX getaddrinfo/getnameinfo as they provide a good top-level interface which - is quite portable (even on Windows) - -Good code examples: -- UNPV12e of W. Richard Stevens (Unix Network Programming), the must-read -- postgresql has a good getaddrinfo replacement for Windows and Cygwin -- some very helpful tips: www_kame_net_newsletter_19980604.txt diff --git a/docs/port/TODOS b/docs/port/TODOS deleted file mode 100644 index ffdb327..0000000 --- a/docs/port/TODOS +++ /dev/null @@ -1,10 +0,0 @@ -- the HAVE_XXX flags could produce a name clash with some autoconf'ed - software/library we will be using in production code. Rename to - WOLF_SYS_HAVE_XXX or similar -- feedback to the c99 snprintf guy, errors like: - - -1 instead of buffer size on boffer overflow - - extern errno in mutli-threaded code -- add a lot more tests -- %zd and similar pointers: make format macros like PRIdPTR which - get mapped to %ld. But what if snprintf could do a %zd and - fprintf not? diff --git a/docs/service/TODOS b/docs/service/TODOS deleted file mode 100644 index a67992a..0000000 --- a/docs/service/TODOS +++ /dev/null @@ -1,6 +0,0 @@ -- option parsing on windows? -- have a generic message/po-file extractor, the perl script here is - an evil copy! -- handle pause/continue -- test long-duration startup and shutdown phases with checkpoint_hints - -- cgit v1.2.3-54-g00ecf