[CWB] Problems with CQPweb frequency list permissions

[Hardie, Andrew]

Hi Philipp,

Aha! That (the collocation discount applying to other things) is a bug I fixed in 3.3. I’ve just backported the fix to 3.2.43 in the repo.

best

Andrew.

From: cwb-bounces at sslmit.unibo.it <cwb-bounces at sslmit.unibo.it> On Behalf Of Philipp Heinrich
Sent: 08 April 2022 11:11
To: Open source development of the Corpus WorkBench <cwb at sslmit.unibo.it>
Cc: Fabian Schäfer <fabian.schaefer at fau.de>
Subject: Re: [CWB] Problems with CQPweb frequency list permissions


Thanks, Stephanie, for your help and to Andrew for his explanations.

I already thought that there might be a difference between privileges for frequency list creation and other ad hoc databases, but I (as a more or less sane person) had not searched at the bottom of the user page for further privileges. Instead I had commented out the relevant bit in db-lib.php

#       if ($num_of_rows > $table_max)
#       {
#               unregister_db_process();
#               exiterror("The action you have requested uses up a lot of diskspace.\n"
#                       . "Your limit is currently set to $table_max instances.\n"
#                       . "Please contact your system administrator if you need access to the information you requested."
#                       );
#       }

which did the job. (Or rather, this was done by a more daring colleague of mine.)

It seems to me though that – at least in the outdated v3.2.41 running on our server – the $table_max value is effectively divided by four also for frequency breakdowns.

This is bizarre because (a) a breakdown of 300,000 matches should have less than 1 mio. instances and (b) the user has been explicitly granted the right to build frequency lists of up to 1 billion tokens.



The 1,000,000 limit should allow you to do a frequency breakdown of 300,000 matches, but for a collocation database the # of matches is multiplied by a (default) cost factor of 4, giving an effective table size of 1,200,000 > 1,000,000 rows.



With the value being set to 1,000,000, I can create frequency breakdowns for a query with 241,233 matches – but not for one with 255,249 matches.

Thanks again and all the best,

Philipp
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://liste.sslmit.unibo.it/pipermail/cwb/attachments/20220408/c9e5e1a6/attachment-0001.html>