Age | Commit message (Collapse) | Author |
|
replying,
forwarding and bouncing.
|
|
|
|
|
|
|
|
counts users of Alpine. This eliminates any information sent to the
developers of Alpine by Alpine users. Based on work by Geoffrey Thomas.
|
|
fixed those that belong to the source code of Alpine and do not come
from an external source. List contributed by Jens Schleusener.
|
|
|
|
of Openssl that do different things. I need to understand them
better. In the mean time, we go back to the original code.
|
|
consistent in what it signs. Problem reported by Björn Krellner.
|
|
fails to add a trailing line to signed text. Reported by Stefan Mueller.
|
|
because of that, Alpine sets their encoding to quoted-printable, in
contradiction to RFC 2045. We handle this by keeping the encoding
reported by the server. Based on a report by Holger Trapp.
|
|
released by the system.
|
|
and short-lived until release of Pine 4.43 (released November 28,
2001.) Also update message to mean what Alpine is trying to do
(forwarding, not answering.)
|
|
removing parameters of the content-type header.
|
|
then forwarding the message will set the first part to be flowed, and
sent that way even when the option Do Not Send Flowed Text is
enabled.
To avoid this issue, we remove any flowed text parameters from the
body of the message, so we create a "remove_parameter" function, and
then we set the flowed parameter only when the user requested.
Reported by Holger Trapp.
|
|
closed.
* A message could fail to verify its signature even if the
certificate was saved when the message was open. Based on a report by
David Woodhouse to the RedHat bugzilla system.
|
|
* Reversing the clearing of the rmap memory.
|
|
us.
* Some distributions of OpenSSL do not define ERR_free_strings and
EVP_cleanup, so we only execute them when defined. Reported by Erich
Eckner.
|
|
calendar.
When "More Details" is selected for each individual event, the description
of that entry is shown only. When the calendar is opened from the view
attachment list screen, all entries will be shown. Mske sure there are no
memory leaks.
* Reverse some of the changes to make valgrind happy, as this may make Alpine crash
by using memory that was released.
|
|
|
|
* rewrite of some code in pith/ical.c to make sure function prototypes are
correct.
* Add a separator line to the calendar entry in case, the calendar is sent
as the main body of the message.
|
|
|
|
|
|
* Update of copyright notice
* Update to release notes to indicate support of RFC 2971.
|
|
|
|
* Add command line argument -smimedir, which allows to specify
the default path for a directory that contains the public, private,
and ca directories. This is useful in case a user has a backup of
old certificates that cannot be installed in the ~/.alpine-smime
dir.
* Update to alpine man page to include documentation on missing command
line options such as -nowrite_password_cache, -passfile, -pwdcertdir,
and -smimedir.
* Various changes in the code to quell some compiler issued warnings in
Mac OSX. Reported by Joe St Sauver. This includes the removing
deprecated ldap functions from the code and the test in the configure
script. The switch to not deprecated functions is done by the use of
the belvar structure, which is not completely appropriate for what we
are doing, but it is sufficient for our needs. The berval structure
is more appropriate for binary data, but it works well with string
data, which is what we need.
* Various changes in the code to quell some warnings issued by clang
3.5.
|
|
space before folding the subject line.
|
|
* Further enhancement to the configure script in finding the location
of the SSL include and library files, when they are installed in th e
default location by openssl.
* When Alpine sends an attachment, it will set the boundary attribute
in lower case, as some SMTP servers, such as those of libero.it
reject messages if the boundary attribute is in uppercase.
* Alpine fails to remove temporary files used during a display or sending
filter. Fix contributed by Phil Brooke.
* SMIME: Crash when checking the signature of a message that contains
a RFC822 attached message. Reported by Holger Trapp and Bjorn Krellner.
|
|
* crash on importing certificates that do not have an email address
associated to them, such as those of a Certificate Authority.
* Disable saving new passwords to the password file. Implemented
by Louis Raphael from dpslabs.com.
* Panda IMAP does not decode correctly Korean text encoded in UTF-8.
Reported by Chulho Yang.
|
|
* Aggregate operations allows bouncing a list of messages using a role.
Suggested by Ulf-Dietrich Braumann.
* Compilation error of module pith/reply.c if SMIME is not defined (as in
Windows Alpine). There was a misplaced parenthesis.
* Update to S/MIME to explain how to use a PKCS12 certificate in
Alpine.
* Fix error in compare_certs function, that would modify the name of
the certificates after sorting them, and return when no certificates
are given.
* When replying to several messages, subject will be decoded first,
and then stripped from re/fwd before they are compared to determine
the subject of the replied message.
* Add $(LIBINTL) to the flags to link rpdump, rpload, alpined and
alpineldap because MAC OSX 10.8 x86_64 needs it.
* When the download of an attachment is interrumpted, Alpine stills
caches what was downloaded, making the download incomplete for
subsequent calls of Alpine attempting to open the attachment. In the
future, Alpine will not cache any downloaded part of the attachment
when it is interrupted.
|
|
* Alpine would not parse options from the command line, such
as -patterns-filters2, correctly.
* Add /usr/local/include as a path to find include and libs files
for openssl in FreeBSD.
* Management certificate screen now prints, in addition to the e-mail
address of the owner of the certificates, the dates of validity
and the MD5 hash of such certificates.
* crash when processing message/rfc822 attachments that are encoded
in base64.
* Openssl: if /usr/local/ssl exists, assume that this is the intended
place where ssl libraries, include files and certificates are located.
Typically, distributions do not use this directory, so its existence
indicates that Openssl has been specially installed there, so it
is probably a preferred place to get the system Openssl files.
* Postponed messages whose content-type is text/html, text/enriched and
text/richtext are sent with that content-type, even though, after
resuming composition, Alpine had changed its type to text/plain.
* HTML: <BR>, <BR />, and <BR/&> are considered the same inline tag;
the same is valid for the <HR> tag.
|
|
* S/MIME Alpine would compute incorrectly the signature of a message
that contains 8bit if the option "Enable 8bit ESMTP Negotiation" is
enabled, the message contains 8bit characters and the smtp server
supports 8bit sending.
* Crash while redrawing S/MIME configuration screen when importing a
certificate
* When forwarding a message before opening it, the message might not be
found. The problem is in the forward_body function, where the section
of the body is not correctly set in all instances.
* When forwarding a signed message Alpine might forward the message as
a multipart message, instead of just selecting the body of the
message. Change to forward the signed part only. This aligns Alpine
with what it does when it replies to a similar message.
|
|
* Introduce the option "Validate Using Certificate Store Only", and
make it the default. This will make Alpine check for the validity of
signatures in certificates that a user has installed in their
system, and not in the certificates that come with the message. A
user can override this, although is not recommended, by disabling
this feature.
* When viewing a signed message, the ^E command would present an
empty screen or Alpine would crash because when Alpine would get the
PKCS7 body of the message from body->sparep, it would not decode it
properly due to the new way in which the sparep pointer is encoded
that was introduced in version 2.19.991.
* When a signed message is forwarded, the message might not be filtered
correctly, and mime information might make it into the body of the
forwarded message. In order to produce this, the message must be
forwarded from the index screen and not be opened. The reason why
this makes a difference is because opening a signed message changes
its body structure. The reason why a person could forward a message
before reading it is because the person could already be aware of
the content of the message (e.g. the message is in the sent-mail
folder).
* When a message fails to validate and the body is saved from the server
for validation, be careful in the way that body part pointers are
set, in order to do this we split the mail_body function into two
parts, one that gets the body, the other that gets the section of
the body. The new function that gets the section of the body
(mail_body_section), is used to assign pointers of the reconstructed
new body.
* When a container has not been defined, transferring messages to a
container will succeed, and the name of the container will be
written on screen.
* When Alpine is receiving the envelopes from an imap server, it
attempts to generate the index line immediately; while doing so it
might need to compute a score, and for this, it might need to go
back and do some operation in the same imap server. In this case,
Alpine will crash with a "lock when already locked" message. In
order to avoid this crash, a new check in match_pattern was added to
Alpine to avoid the second trip to a server that is busy sending us
envelopes. Reported by Peter Koellner.
* Update copyright notice in mswin.rc and pmapi.rc, as well as first
time user notice and special request notice.
* Alpine cannot handle correctly some characters in the Windows-1256
character set, which might lead to a crash or a corruption in the
screen. Work was done to contain the bug. A more complete fix will be
done in a future release. Reported by Professor Robert Funnell.
* Decode the name of attachment names, so they can be written as part
of the description of the part.
* When transferring certificates to a local container, create container
with default names PublicContainer, PrivateContainer and CAContainer,
as appropriate for these files, unless the user has provided some
other names.
|
|
works.
* When a message is sent encrypted, add the sender certificate so that
the sender can decrypt it too.
* When a message is signed and encrypted, first sign it and then encrypt
it. This changes the usual order of encrypting and then signing, and it has
the shortcoming of making bigger messages. However, this is the way that
most clients work with S/MIME, and so for compatibility with other programs,
we will send signed, then encrypted, instead of encrypted, then signed.
Hmm... should we sign the encrypted part?
* Avoid the first RSET smtp command, as this causes delays in some evily
managed servers.
|
|
body->nested.part instead of b->nested.part in the
do_detached_signature_verify function, and save its body and mime
headers in create_local_cache. Now all signed messages should
verify correctly.
* Protect against potential crash bug in write_passfile function by
checking if text != NULL. text can only be null if there are no
passwords to save. We assume we could get write_passfile called
with null arguments, so this is just to protect that.
* Add handling of corner cases to several functions by initializing some
variables. Reported and patched by James Jerkins.
* When selecting the certificate/key pair to encrypt/decrypt the password
file choose it in this order:
- if -pwdcertdir is given look for certificates there, if nothing
there, we bail out;
- otherwise we look in the default directory, if anything there
and it matches to be a key/cert pair, we use it;
- otherwise we check if smime_init() has been called. If not
we call it;
- we check if a key/cert pair has been found with smime initialized,
if so, use it and copy it to the default directory;
- if not, check if there is anything in the default smime directories
(.alpine-smime/private and .alpine-smime/public), and in this
case copy it to the default.
- otherwise we bail. We will eventually create a certificate/key
pair for the user;
- finally, if we called smime_init(), we call smime_deinit().
Throughout this process, if smime_init() was not called before
we tried to get the cert/key pair we exit this process without
ps_global->smime->inited set, so that other process that need
to call smime_init() get the right structure initialized. This is
done because we might pick a cert/key pair to decrypt the password
file.
|
|
for patches.freeiz.com/alpine/. Work in progress.
* prototype function tigetstr in pico/osdep/terminal.c
* folders encoded in modified utf7 are transformed their names to a human
readable utf8.
* New attempt to fix smime support in Alpine. Messages sent by alpine
with or without attachments should validate in all servers and in
all folder formats.
|
|
* check bounds and tie strings off to improve security. Contributed by
James Jerkins.
* Alpine crashed when a user attempted to add a folder collection, due
to bug in GET_NAMESPACE in imap4r1.c.
|
|
* Alpine computes the signature of a signed message on a different message than the one
it sends, because of this, messages that are sent can not be validated either in the
sent-mail folder, nor in the server that receives such message.
|
|
non-compliant servers do not understand uppercase content-type, such as
those of GMX.de.
* Transformation of UTF-8 to MUTF7 was not being done when creating a folder
in an IMAP server.
|
|
* Pico: search backwards.
* do not allow List- headers to be set up by users in accordance with
RFC 2369.
* minor fixes for new justification support (add comparison of a line
with the previous *and* the next).
|
|
|
|
a url by
a browser.
|
|
|