without a comparison operator, it tests for the field's presence or
absence, not its value; to test whether a Boolean field is true, you
compare it with a non-zero value, and to test whether it's false, you
compare it with a zero value.
Make the filter expression construction dialog handle that correctly.
svn path=/trunk/; revision=3068
always attached to the dialog as the E_FILT_FILTER_TE_KEY data, but only
sometimes attached as the E_FILT_TE_KEY data.
Get rid of E_FILT_TE_KEY completely, as it's redundant, and use only
E_FILT_FILTER_TE_KEY; this keeps us from crashing as a result of trying
to manipulate the widget referred to by E_FILT_TE_KEY if E_FILT_TE_KEY
hasn't been set to refer to any widget.
svn path=/trunk/; revision=3067
and, when it's being destroyed, disconnect from the "destroy" signal on
the text entry box to which it's attached, so that, when that text entry
box is destroyed, we don't try to get rid of the no-longer-extant
filter-expression-construction dialog.
svn path=/trunk/; revision=3061
filter-expression-construction dialog box is attached; if the text entry
box is destroyed (which typically means the window it's in was
destroyed), get rid of the filter-expression-construction dialog box.
svn path=/trunk/; revision=3060
available for it and it looks ugly" is "throw an alignment around it".
(I *still* don't know why it's not required in other dialog boxes, e.g.
the filter-editing dialog box.)
svn path=/trunk/; revision=3059
expression construction dialog activate the entire dialog box.
Make a desperate but failed attempt to bludgeon GTK+, The Toolkit That
Knows Better Than You Do How Big Buttons Should Be Made, Even If It
Looks Butt-Ugly, And Which Appears To Randomly Decide Whether To Make It
Look Ugly Or Not, into making the "Cancel" button as tall as the inside
of the "Accept" button, not as tall as the "Accept" button plus its
"this is the default button" surround.
svn path=/trunk/; revision=3057
fields.
Show a relational operator on a field if the field supports it *or* if
the field can be sliced and the type generated by a range (FT_BYTES)
supports it. (This lets you do a comparison on a protocol, not just on
a range of a protocol - e.g., "arp == 2", not just "arp[0:1] == 2" - but
the alternative would be to show only the "is present" test, as if you
don't offer the other tests, you can't turn on the range text box when
they select a comparison expression if you don't show comparison
expresions...).
svn path=/trunk/; revision=3056
anything from the packet, so that if we throw an exception before
setting the Info column, it doesn't have stuff left over from the
previous dissector.
Tvbuffify the RADIUS dissector.
svn path=/trunk/; revision=3055
and deleting it might've confused the stuff that generates the anonymous
CVS tree, as it didn't propagate the deletion. I'll delete it again
after this checkin shows up in the anonymous CVS tree.
svn path=/trunk/; revision=3054
application, catch what GLib message-logging calls we can, and create a
console and make it the standard input, output, and error if such a call
is made, so those messages show up in a console window. Create the
console for the output of "ethereal -v" as well.
svn path=/trunk/; revision=3052
defined.
Use "gint64" and "guint64", not "long long int", for 64-bit integral
types, so that this code works with compilers (such as Microsoft Visual
C++) that have 64-bit integral types but that don't call them "long
long".
Use "pntohll()" to extract 64-bit integral types from a field.
Put a "break;" into a "default:" clause - MSVC++ doesn't like
switch (XXX) {
...
default:
}
svn path=/trunk/; revision=3051
It doesn't do DIAMETER-over-UDP, so the RADIUS dissector no longer
checks for DIAMETER packets and calls the DIAMETER dissector if it finds
one.
svn path=/trunk/; revision=3049
expect - it means "same as fully-buffered". This means that the "-l"
flag is a no-op on Windows.
Instead of setting line-buffered mode with "setvbuf()", set a flag and,
if that flag is set, flush the standard output after the information for
ever packet is printed; this isn't "line-buffered", either, but, as the
reason for doing line-buffering is to allow the output of Tethereal to
be piped to a program and to have that program see the output for a
packet as soon as the packet is seen and dissected, it should be just as
good as line-buffered.
svn path=/trunk/; revision=3047
the function have the expected type.
Make the arguments to comparison functions used by the merge sort be
"const void *", not "void *", just as the arguments to the comparison
functions used by "qsort()" are "const void *".
svn path=/trunk/; revision=3040
take fully-prototyped function arguments with types appropriate to
"g_malloc()" and "g_free()", and change the calls to the functions
pointed to by those arguments not pass the extra __FILE__ and __LINE__
arguments.
svn path=/trunk/; revision=3039
LP64 platforms.
Change "lemon" to use function prototypes and the official ANSI C style
of variable-argument-list functions, and to include various system
header files rather than to use non-prototype declaration of various
system functions, to do a lot more type checking.
svn path=/trunk/; revision=3037
"pinfo->current_proto" - this routine is called only through a dissector
table, and the code to call through a dissector table does that stuff
for you.
Clear the Info column before doing anything that could throw an
execption, so that if an exception is thrown the display doesn't show
junk left over from the protocol above us.
Get rid of the GCCism "case N ... M".
svn path=/trunk/; revision=3036
- it only allows you to convert a *signed* 64-bit integer to a "double".
Cast the result of "pletohll()" to "gint64" before returning it from a
function that returns a "double".
svn path=/trunk/; revision=3033
in some places use "guint64", on plaforms where it's available,
rather than floating point (we don't yet use it universally, as
we'd have to provide code to do 64-bit arithmetic on
platforms/compilers where 64-bit integral types aren't
supported);
use .838096 microseconds rather than 1 microseconds as the time
stamp units for NetXRay 2.x format, as those capture files seem
to use that time stamp (that's the Sniffer "PC" time stamp;
perhaps when Network Associates assimilated Cinco, they changed
the time stamp units).
svn path=/trunk/; revision=3027
in the output of "{ethereal,tethereal} -G", so that it appears only once
in the documentation.
Expand some comments to give more details.
svn path=/trunk/; revision=3024
can now handle that; this allows us to register both the modulo-8 and
the modulo-128 versions of various X.25 bitfields with "x.25.XXX" names,
which lets us get rid of the "ex.25" protocol stuff completely and use
"x.25" for both modulo-8 and modulo-128 X.25. Do so. (Also, fix up
some cases where we appeared to be using the modulo-8 fields when
dissecting modulo-128 X.25.)
This, in turn, allows us to register the X.25 dissector, as there's now
only one protocol with which it's associated, and make it static and
have it called only through a handle, and to, when registering it with
the "llc.dsap" dissector table, associate it with "proto_x25".
That, in turn, allows us to get rid of the "CHECK_DISPLAY_AS_DATA()"
calls, and the code to set "pinfo->current_proto", in the X.25
dissector.
The code for the display filter expression dialog would, if there are
two fields with the same name registered under a protocol, list both of
them; have it list only one of them - the fields should have the same
type, the same radix, and the same value_string/true_false_string table
if any (if they don't, they're really not the same field...).
svn path=/trunk/; revision=3023
Guy Harris