select varchar - thinks the package is incomplete while it's actually complete

[simendsjo]

EDIT: I wrote the wrong compiler version

I'm not really sure what is causing this as I no longer remember the protocol.
This is running on linux x64, mariadb 5.5.32, dmd 2.063.2.

Could anyone test this on mysql? It's up for grabs if anyone wants it :)

    auto db = new MysqlDB("host=localhost;port=3306;user=test;pwd=test;db=homepage");
    auto cn = db.lockConnection;
    scope(exit) cn.close();
    auto cmd = new Command(cn, "create table if not exists mysqlnbug (s varchar(64));");
    ulong n;
    cmd.execSQL(n);
    cmd.sql = "select s from mysqlnbug";
    // In consumeNonBinaryIfComplete!string, isIncomplete=true because LCB value > packet.length
    // this causes Row to continue fetching packets, but there are no more data from
    // the server, so this just hangs forever.
    cmd.execSQLResult();

[sshamov]

I cannot test the case right now, but I noticed that a packet is not checked by calling packet.isEOFPacket() inside do-while loop at Row constructor. Maybe this is the problem.

[punkUser]

I'm seeing this issue here too. Field is a varchar(256) in my case and if it's sufficiently long the query will just hang within the loop. Any fixes or ideas what to do to work around?

[Abscissa]

@simendsjo: Your sample code doesn't reproduce the issue for me even if I insert a row with the maximum length, 64 bytes.

However, it does hang for me if I change it to varchar(256) and insert a row with at least 251 bytes. Is that perhaps what you meant? If so, then it should be fixed now, via #40. Can you verify?

[Abscissa]

Pinging @simendsjo: Since I (still) can't reproduce the exact original problem with varchar(64), but can verify that #40 does fix it for varchar(256) and a row with >=251 bytes, I'm going to go ahead and assume this issue is fixed and close the ticket. Please re-open if the problem still occurs.