This is programmatically generated dot code, but even simplified, it still doesn't leave empty cells for the even-numbered ports - I'm expecting alternating value/empty cell with pointer pairs. I tried adding spaces, but the parser is apparently just dumping them.
I believe the problem is somewhere in the port-string parsing, but I'm
not a good-enough C programmer to spot the problem immediately. If I
manage to come up with a patch, I will let you know. If you patch this
one earlier, please let me know.
Input:
Comments:
digraph "test" {
"gvds_array0" [
label =
"{<port1>Nil}|{<port2>}|{<port3>Nada}|{<port4>}|{<port5>Zip}|{<port6>}"
shape = "record"
]
"gvds_array1" [
label = "[]"
shape = "plaintext"
]
"gvds_array2" [
label = "[]"
shape = "plaintext"
]
"gvds_array3" [
label = "[]"
shape = "plaintext"
]
{
"gvds_array0"
}
{
"gvds_array1"
"gvds_array2"
"gvds_array3"
}
"gvds_array0":"port2" -> "gvds_array1" [
]
"gvds_array0":"port4" -> "gvds_array2" [
]
"gvds_array0":"port6" -> "gvds_array3" [
]
}
[erg] Empty field labels are changed into " " in parse_reclbl, but
in storeline, both CodeGen->textsize and estimate_textsize assign
this text 0 width. Thus, the field disappears. Previously, actualwidth
would always give this some non-zero width. Clearly, some special case
code somewhere needs to be added to do handle this, as well as whatever
is expected using hard spaces.
Owner: ellson
Status: Fixed (17 Jan 2002)