- Event Series
- This event is an instance of E422: Weekly Standup, and repeats every week.
Event Timeline
Comment Actions
- Today's topic: libdns T4367: gnupg fails to build with gcc9:
- Two things for portability
- C99 feature of struct/union initializer specifying member
- (non portable, even with C99) compound literals as static constant, example is: &(struct timespec){ 0, 0 }
- In my opinion, the former would be OK to keep, but latter is better to be removed
- If agreed, I'd like to remove this
- a reference: https://stackoverflow.com/questions/31863490/static-struct-initialization-in-c99
- Two things for portability
- Last week:
- Fixed
- T2011: gnupg should notify cancellation of its operation to gpg-agent to kill pinentry
- by spawning a thread watching socket (of gpg<->gpg-agent) when GETPIN is requested to pinentry
- Backport some changes in master (in the period of 2.2.0 - 2.2.2) which has been in Debian to sync Debian distribution of GnuPG 2.2
- disable scdaemon when 'make check' for master: rG64b7c6fd1945: tests: Add "disable-scdaemon" in gpg-agent.conf.
- (before this change) when we insert Gnuk Token and it is unused, 'make check' may fail
- Fix handling of user interaction timeout (ack button) in master: rGbd15aa34ab8a: scd: Handle ack button timeout as GPG_ERR_TIMEOUT., rG2013cb5ee667: scd: Clear CHV status on timeout error.
- Possibly important fix (will be in 2.2?): rG611faf1579a5: scd: internal driver: Submit SET_INTERFACE control transfer.
- USB (in general) doesn't have concept of "session" by open/close
- while its API has claim/release to allow a single program/driver's exclusive access
- But with UIF feature, when a user cancels the operation by killing the process (of scdaemon), an endpoint keeps data sending (from device), this should be cleaned up, when the device will be "opened" again. Unfortunately, cleanup at exiting is not reliable.
- It is only by internal driver. PC/SC driver does nothing like this.
- So, a user has to remove/reinsert the device, after killing
- USB (in general) doesn't have concept of "session" by open/close
- T2011: gnupg should notify cancellation of its operation to gpg-agent to kill pinentry
- Fixed
- This week:
- libdns portability issue
- Gnuk supports of timeout, possibly new release
- follow-up Damien's change of libksba
- smartcard access with Chopstx
Comment Actions
Last week:
- Scute work: Signing and decryption using a PIV card does now work with Thunderbird.
- Changed Scute license to LGPLv2.1+
- Updated build system to current standard. Can again be build for Windows.
This week:
- Fix recently found bugs in Scute.
Comment Actions
Last week:
- Updated KDE dependencies and Qt in Gpg4win
- Many patches could be removed because KDE supports more reduced builds (e.g. without dbus for android)
- Parsed autocrypt header in GpgOL
- Added simple "encrypted subject" read support to GpgOL
- Some bugfixes for GpgOL
- Made "binary icon resource" work for cross compiling to include Kleopatra's icons in the binary.
This week:
- Add scute to Gpg4win
- Have another look at the gpgtar / kleopatra archive corruption.
- In general -> prepare Gpg4win-3.1.6