aboutsummaryrefslogtreecommitdiffstats
path: root/common/estream.c
diff options
context:
space:
mode:
authorWerner Koch <[email protected]>2011-02-08 20:11:19 +0000
committerWerner Koch <[email protected]>2011-02-08 20:11:19 +0000
commit2c79a2832cd3cbc1c09f4f7a7b2653ba6cbd2845 (patch)
tree87e424f10a74c0cecce3c3ecc96b43eaca1b892c /common/estream.c
parentFix ECDSA 521 bit signing. (diff)
downloadgnupg-2c79a2832cd3cbc1c09f4f7a7b2653ba6cbd2845.tar.gz
gnupg-2c79a2832cd3cbc1c09f4f7a7b2653ba6cbd2845.zip
Add finger support to dirmngr.
The basic network code from http.c is used for finger. This keeps the network related code at one place and we are able to use the somewhat matured code form http.c. Unfortunately I had to enhance the http code for more robustness and probably introduced new bugs. Test this code using gpg --fetch-key finger:[email protected] (I might be the last user of finger ;-)
Diffstat (limited to 'common/estream.c')
-rw-r--r--common/estream.c11
1 files changed, 8 insertions, 3 deletions
diff --git a/common/estream.c b/common/estream.c
index bc820513e..a73d1f2c4 100644
--- a/common/estream.c
+++ b/common/estream.c
@@ -3041,9 +3041,14 @@ es_fclose (estream_t stream)
already registered notification; for this to work the value of FNC
and FNC_VALUE must be the same as with the registration and
FNC_VALUE must be a unique value. No error will be returned if
- MODE is 0. Unregistered should only be used in the error case
- because it may not remove memory internall allocated for the
- onclose handler.
+ MODE is 0.
+
+ FIXME: I think the next comment is not anymore correct:
+ Unregister should only be used in the error case because it may not
+ be able to remove memory internally allocated for the onclose
+ handler.
+
+ FIXME: Unregister is not thread safe.
The notification will be called right before the stream is closed.
It may not call any estream function for STREAM, neither direct nor