[bug #60246] When writing stdout into a pipe, wget never terminates

[Darshit Shah]

Update of bug #60246 (project wget):

                Severity:              3 - Normal => 4 - Important          
                  Status:               Confirmed => In Progress            
             Assigned to:                    None => darnir                 

    _______________________________________________________

Follow-up Comment #2:

Okay, this is a crazy issue, and I'm surprised no one has reported it in so
many years! I think this issue has likely existed for 15 years if not more.

The problem is simple:
We pass the file descriptor of the file to which the downloaded document is
saved in order to parse it for any links. So, when you combine -p with -O-,
what you get is that Wget waits indefinitely on a read() call on stdout. Which
will _NEVER_ return anything.

In order to handle -p and -O properly, I think we should download to a
temporary file before parsing and then append it to the output file. This is
especially important when using -O-. When writing to a real file, we could
probably optimize the process by writing directly to the file and then calling
fseek() to only process the newly downloaded part of the file.

I would wager this is also an issue with -r -O-. 

    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?60246>

_______________________________________________
  Message sent via Savannah
  https://savannah.gnu.org/