Tcl Source Code

View Ticket
Login
Ticket UUID: 2953593
Title: [gets $fd data] returns 0 in nonblocking mode
Type: Bug Created on: 2010-02-17 15:12:52
Submitter: egavilan Assigned to: andreas_kupries
Subsystem: 24. Channel Commands Severity:
Priority: 5 Medium Last modified: 2026-10-22 17:19:27
Status: Closed Closed by: serhiy.storchaka
Resolution: Fixed Closed on: 2026-10-16 21:44:20
Version: None
Description:
Reading a serial port in nonblocking mode sometimes returns 0 for the number of bytes read.

The [gets] documentation states "If varName is specified and an empty string is returned in varName because of end-of-file or because of insufficient data in non-blocking mode, then the return count is -1".

In this case I expect either -1 or 10 (the length of the incoming message from the device, which never sends empty lines).

The next script shows the issue:

===========================================================================
proc readserial {fd} {
    global short empty normal

    set chars [chan gets $fd data]
    if {$chars == -1} {
        if {[chan eof $fd]} {
            chan close $fd
            done
        }
        incr short
        return
    }
    # this shouldn't happen, but it does anyway
    if {$chars == 0} {
        incr empty
        return
    }
    incr normal
    if {$normal == 10000} {
        done
    }
}

proc done {} {
    global short empty
    chan puts "Empty reads: $empty\nShort reads: $short"
    exit
}

set empty  0
set short  0
set normal 0

set fd [open /dev/ttyUSB0 r]
chan configure $fd \
    -mode 9600,n,8,1 \
    -buffering line \
    -blocking 0
chan event $fd readable [list readserial $fd]
===========================================================================

Notes:
In blocking mode there's no "empty" reads.
This happen in both windows and linux.
The serial port is a usb<->serial adapter.
The whole script has a Tk gui, showing the value read in a label.
User Comments:
ferrieux added on 2010-03-02 21:37:30:
strace appreciated ;-)

serhiy.storchaka added on 2026-10-16 21:40:49:

Not reproduced: CRLF, LF and CR input split at random points gives no unexpected empty lines (LF CR input gives an empty line for each CR in auto mode, as expected). Testing this found a regression in 8.6.13, reported as [4c5a98f5a0].


serhiy.storchaka added on 2026-10-16 21:44:20:

Correction: this is the torn CRLF case of [b3977d199b] (CR and LF in different reads give an extra empty line in auto mode). It reproduces in 8.4.20 to 8.6.12 and was fixed in 8.6.13 by [a3ed092aab]. That check-in also caused [4c5a98f5a0].