gpgme/TODO

Hey Emacs, this is -*- outline -*- mode!

* IMPORTANT
** When using descriptor passing, we need to set the fd to blocking before
   issueing simple commands, because we are mixing synchronous
   commands into potentially asynchronous operations.
** Might want to implement nonblock for w32 native backend!  Right now,
   we block reading the next line with assuan.

* Before release:
** Figure out if _gpgme_io_pipe should pre-create reader/writer and if we
   then can use !start_it in most invocations.  Note that gpgme_io_dup
   probably should always use !start_it, because currently it will
   always create reader/writer if they don't exist(!?).
** Some gpg tests fail with gpg 1.3.4-cvs (gpg/t-keylist-sig)
   The test is currently disabled there and in gpg/t-import.
** When gpg supports it, write binary subpackets directly,
   and parse SUBPACKET status lines.
** A few months after 1.1.7:
   Remove GPGME_CONF_PATHNAME macro.

* ABI's to break:
** gpgme_edit_cb_t: Add "processed" return argument
   (see edit.c::command_handler).
** I/O and User Data could be made extensible.  But this can be done
   without breaking the ABI hopefully.
*  All enums that should be enums need to have a maximum value to ensure
   a certain minimum width for extensibility.
** Compatibility interfaces that can be removed in future versions:
*** gpgme_data_new_from_filepart
*** gpgme_data_new_from_file
*** gpgme_data_new_with_read_cb
*** gpgme_data_rewind
*** gpgme_op_import_ext
*** gpgme_get_sig_key
*** gpgme_get_sig_ulong_attr
*** gpgme_get_sig_string_attr
*** GPGME_SIG_STAT_*
*** gpgme_get_sig_status
*** gpgme_trust_item_release
*** gpgme_trust_item_get_string_attr
*** gpgme_trust_item_get_ulong_attr
*** gpgme_attr_t
*** All Gpgme* typedefs.


* Thread support:
** When GNU Pth supports sendmsg/recvmsg, wrap them properly.
** Without timegm (3) support our ISO time parser is not thread safe.
   There is a configure time warning, though.

* New features:
** Flow control for data objects.
   Currently, gpgme_data_t objects are assumed to be blocking.  To
   break this assumption, we need either (A) a way for an user I/O
   callback to store the current operation in a continuation that can
   be resumed later.  While the continuation exists, file descriptors
   associated with this operation must be removed from their
   respective event loop.  or (B) a way for gpgme data objects to be
   associated with a waitable object, that can be registered with the
   user event loop.  Neither is particularly simple.
** Extended notation support.  When gpg supports arbitrary binary
   notation data, provide a user interface for that.
** notification system
   We need a simple notification system, probably a simple callback
   with a string and some optional arguments.  This is for example
   required to notify an application of a changed smartcard, The
   application can then do whatever is required.  There are other
   usages too.  This notfication system should be independent of any
   contextes of course.
** --learn-code support
   This might be integrated with import. we still need to work out how
   to learn a card when gpg and gpgsm have support for smartcards.
** Might need a stat() for data objects and use it for length param to gpg.
** Implement support for photo ids.
** Allow selection of subkeys
** Allow to return time stamps in ISO format
  This allows us to handle years later than 2037 properly.  With the
  time_t interface they are all mapped to 2037-12-31
** New features requested by our dear users, but rejected or left for
   later consideration:
*** Allow to export secret keys.
    Rejected because this is conceptually flawed.  Secret keys on a
    smart card can not be exported, for example.
*** Selecting the key ring, setting the version or comment in output.
    Rejected because the naive implementation is engine specific, the
    configuration is part of the engine's configuration or readily
    worked around in a different way
*** Selecting the symmetric cipher.
*** Exchanging keys with key servers.

* Documentation
** Document validity and trust issues.
** In gpgme.texi: Register callbacks under the right letter in the index.

* Engines
** Do not create/destroy engines, but create engine and then reset it.
   Internally the reset operation still spawns a new engine process,
   but this can be replaced with a reset later.  Also, be very sure to
   release everything properly at a reset and at an error.  Think hard
   about where to guarantee what (ie, what happens if start fails, are
   the fds unregistered immediately - i think so?)
   Note that we need support in gpgsm to set include-certs to default
   as RESET does not reset it.
** Optimize the case where a data object has 0an underlying fd we can pass
   directly to the engine.  This will be automatic with socket I/O and
   descriptor passing.
** Move code common to all engines up from gpg to engine.
** engine operations can return General Error on unknown protocol
   (it's an internal error, as select_protocol checks already).
** When server mode is implemented properly, more care has to be taken to
   release all resources on error (for example to free assuan_cmd).
* GPG breakage:
** gpg 1.4.2 lacks error reporting if sign/encrypt with revoked key.
** gpg 1.4.2 does crappy error reporting (namely none at all) when
   smart card is missing for sign operation:
    [GNUPG:] CARDCTRL 4
    gpg: selecting openpgp failed: ec=6.110
    gpg: signing failed: general error
    [GNUPG:] BEGIN_ENCRYPTION 2 10
    gpg: test: sign+encrypt failed: general error
** Without agent and with wrong passphrase, gpg 1.4.2 enters into an
   infinite loop.
** Use correct argv[0]
   In rungpg.c:build_argv we use
      argv[argc] = strdup ("gpg"); /* argv[0] */
   This should be changed to take the real file name used in account.


* Operations
** Include cert values -2, -1, 0 and 1 should be defined as macros.
** If an operation failed, make sure that the result functions don't return
   corrupt partial information. !!!
   NOTE: The EOF status handler is not called in this case !!!
** Verify must not fail on NODATA premature if auto-key-retrieval failed.
   It should not fail silently if it knows there is an error. !!!
** All operations: Better error reporting. !!
** Export status handler need much more work. !!!
** Import should return a useful error when one happened.
*** Import does not take notice of NODATA status report.
*** When GPGSM does issue IMPORT_OK status reports, make sure to check for
    them in tests/gpgs m/t-import.c.
** Verify can include info about version/algo/class, but currently
   this is only available for gpg, not gpgsm.
** Return ENC_TO output in verify result.  Again, this is not available
   for gpgsm.
** Genkey should return something more useful than General_Error.
** If possible, use --file-setsize to set the file size for proper progress
   callback handling.  Write data interface for file size.
** Optimize the file descriptor list, so the number of open fds is
   always known easily.
** Encryption: It should be verified that the behaviour for partially untrusted
   recipients is correct.
** When GPG issues INV_something for invalid signers, catch them.

* Error Values
** Map ASSUAN/GpgSM ERR error values in a better way than is done now. !!
** Some error values should identify the source more correctly (mostly error
   values derived from status messages).
** In rungpg.c we need to check the version of the engine
   This requires a way to get the cached version number from the
   engine layer.


* Tests
** Write a fake gpg-agent so that we can supply known passphrases to
   gpgsm and setup the configuration files to use the agent.  Without
   this we are testing a currently running gpg-agent which is not a
   clever idea. !
** t-data
*** Test gpgme_data_release_and_get_mem.
*** Test gpgme_data_seek for invalid types.
** t-keylist
   Write a test for ext_keylist.
** Test reading key signatures.

* Debug
** Tracepoints should be added at: Every public interface enter/leave,
   before and in every callback, at major decision points, at every
   internal data point which might easily be observed by the outside
   (system handles).  We also trace handles and I/O support threads in
   the w32 implementation because that's fragile code.
   Files left to do:
   data-fd.c data-mem.c data-stream.c data-user.c debug.c rungpg.c
   engine.c engine-gpgsm.c funopen.c w32-glib-io.c wait.c
   wait-global.c wait-private.c wait-user.c op-support.c decrypt.c
   decrypt-verify.c delete.c edit.c encrypt.c encrypt-sign.c export.c
   genkey.c import.c key.c keylist.c passphrase.c progress.c signers.c
   sig-notation.c trust-item.c trustlist.c verify.c
** Handle malloc and vasprintf errors.  But decide first if they should be
   ignored (and logged with 255?!), or really be assertions. !

* Build suite
** Make sure everything is cleaned correctly (esp. test area).
** Enable AC_CONFIG_MACRO_DIR and bump up autoconf version requirement.
   (To fix "./autogen.sh; ./configure --enable-maintainer-mode; touch
   configure.ac; make").  Currently worked around with ACLOCAL_AMFLAGS???

* Error checking 
** engine-gpgsm, with-validation
   Add error checking some time after releasing a new gpgsm.


Copyright 2004, 2005 g10 Code GmbH

This file is free software; as a special exception the author gives
unlimited permission to copy and/or distribute it, with or without
modifications, as long as this notice is preserved.

This file is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY, to the extent permitted by law; without even the
implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
PURPOSE.