BUG in SEARCH /NUMBERS - VMS

This is a discussion on BUG in SEARCH /NUMBERS - VMS ; Alpha VMS 8.3 >$ search/num $disk2:[000000]indexf.sys mozilla.com;5 > 3 >> This happens to be at record 37073 of the file. (I used TPU to edit indexf.sys and then find MOZILLA.COM;5 and then WHAT LINE) But... > $ search/num $disk2:[000000]indexf.sys mozilla.com ...

+ Reply to Thread
Results 1 to 4 of 4

Thread: BUG in SEARCH /NUMBERS

  1. BUG in SEARCH /NUMBERS

    Alpha VMS 8.3

    >$ search/num $disk2:[000000]indexf.sys mozilla.com;5
    >3
    >>


    This happens to be at record 37073 of the file. (I used TPU to edit
    indexf.sys and then find MOZILLA.COM;5 and then WHAT LINE)


    But...


    > $ search/num $disk2:[000000]indexf.sys mozilla.com
    > 26547 (dc)
    >>.X



    So yesterday, I had scratched my head wondering why /NUMBERS didn't
    work, and wrote it off as a feature that only works on text files. But
    today, I find that it is able to print "line numbers" for records that
    are found earlier in the same file. (I have since created new versiosn
    of that lost ;5 file and they are placed higher up in the indexf.sys).

    I suspect a signed short integer may be in use somewhere resulting in
    the record number above 32k not being printed.

    Interesting that the statistics display at the end (when you do a /STAT)
    is able to provide numbers that are way higher than 32k.

  2. Re: BUG in SEARCH /NUMBERS

    On Oct 29, 4:13 pm, JF Mezei wrote:
    > Alpha VMS 8.3
    >
    > >$ search/num $disk2:[000000]indexf.sys mozilla.com;5
    > >3
    > >>

    >
    > This happens to be at record 37073 of the file. (I used TPU to edit
    > indexf.sys and then find MOZILLA.COM;5 and then WHAT LINE)
    :
    > I suspect a signed short integer may be in use somewhere resulting in
    > the record number above 32k not being printed.


    Nah, that would be too big a bug...

    $ pipe perl -le "print for (1..99999)" | sea/num sys$pipe 0000
    10000 10000
    20000 20000
    30000 30000
    40000 40000
    50000 50000
    60000 60000
    70000 70000
    80000 80000
    90000 90000

    No problem with line numbers > 33K

    Hein.


  3. Re: BUG in SEARCH /NUMBERS

    On Oct 29, 3:13 pm, JF Mezei wrote:
    > Alpha VMS 8.3
    >
    > >$ search/num $disk2:[000000]indexf.sys mozilla.com;5
    > >3
    > >>


    Something else must be going on. Maybe a visual effect with rub-outs?
    Send that output to a file an try again?

    Below is a sample session with record 40,000 in an indexf.sys
    suggesting all works fine in at least 1 case.
    Regards,
    Hein.

    $ dump/bloc=(start=40000,count=1) sys$disk:[000000]indexf.sys
    :
    Virtual block number 40000 (00009C40), 512 (0200) bytes
    :
    00000000 00509A15 02010000 FFFF6428 (d........P..... 000000
    :
    00000014 00000000 00000139 0DB3FFCC �.�.9........... 000040
    30303039 35444345 30364124 4C49414D MAIL$A60ECD59000 000050

    $ searc/number/form=nonull sys$disk:[000000]indexf.sys MAIL
    $A60ECD59000
    40000
    (dP.
    >�9MAIL$A60ECD590005009





  4. Re: BUG in SEARCH /NUMBERS

    Hein RMS van den Heuvel wrote:
    > Something else must be going on. Maybe a visual effect with rub-outs?
    > Send that output to a file an try again?


    You are right. Because DECTERM still doesn't have the capabilities of a
    VT100 (display control characters), it wasn't obvious. But sending it to
    a file shows that while SEARCH converts funny characters to their
    printable name, it doesn't do that for CR, it actually spits it out raw,
    and that causes the cursor to return to column 1 and overwrite the number.

    LF and FF also appear to be emitted raw.

+ Reply to Thread