Release 1.7a is being distributed, as network conditions permit, within
the next 48h. To maximize staff availability in case of unexpected
problem, no program file will be sent before 8am EDT. The release notes
are being mailed under separate cover.
Because of changes in the update installation procedure, release 1.7a
will exceptionally come in two parts: a very small UPD16F shipment,
containing just the new INSTALL processor and causing neither reboot nor
change in version number, and the usual UPD17A shipment. For integrity
reasons, the new installation procedure requires more free space on the
191 minidisk than with previous versions; to successfully load the UPD17A
shipment, you will need at least 1.3M of free space. If the disk fills up
during the installation, the CARD deck will be kept and you will be
notified; just increase the size of the 191 minidisk and/or delete old
LFIX backup files, then issue an INSTALL RELOAD UPD17A command.
A justification for these changes can be found in the release notes under
the heading "New INSTALL processor"; you will probably want to read this
section before installing 1.7a anyway. The second most important section
is the one titled "CRC-based checksumming of DISTRIBUTE jobs", which
explains how to deal with jobs LISTSERV identifies as being corrupted.
These two sections have been ordered before the others for your
convenience, and you are strongly advised to read the release notes
sequentially up to that point before proceeding with the installation. I
apologize for the rambling nature of the release notes, and the probably
countless omissions. We are presently being plagued with the unfortunate
collusion of a heat wave and a new alarm system which prevents most
windows from being opened for more than 30 seconds. The building, being
designed for neither heat-generating terminals nor alarm systems, has no
air conditioner; being however designed for swedish winters, it is very
good at not letting the warmth out during the cooler nights (in spite of
the generous contribution of a pair of drunkards who threw a brick
through one of the windows yesterday at 3am). Anyway it took almost 1
week to produce the release notes, and I have probably abused PUT/GET
from notebooks; if you have any questions, mail them to LSTSRV-L.
Finally, I would like to encourage sites which are going to compile the
code to provide some feedback to the LSTSRV-L list, for the benefit of
other sites which may be considering whether or not to purchase the REXX
compiler. Such figures will however be meaningless if they are based on a
comparison with release 1.6e, rather than interpreted 1.7a, as there have
been numerous performance changes. One way to address this problem is to
deliberately run the interpreted version of 1.7a (with LISTPROC enabled -
see the release notes) for a while before switching to the compiled
version. Another way is to collect statistics from sites without the REXX
compiler on the resource usage of LISTSERV before and after the
installation of 1.7a, and compare them to the figures of "compiled"
sites. In any case, make sure you read the "REXX compiler" section in the
release notes carefully before attempting to compile anything.
Eric
|