Date and Description:
12 Sep 08 CIRCULARS NOW MIME-COMPATIBLE:
The occurance rate of people submitting Circulars with HTML versions
attached in the incoming emails has been dramatically increasing.
The Circulars processing program did not handle MIME -- it just distributed
the entire body of the email (very messy when there is HTML in addition
to the plain-text version).
The program has been modified to handle MIME/attachments/etc.
Now only the plain-text version of submitted Circulars are being distributed.
So you, the submittor, no longer has to deal with disabling HTML-ized versions
in your mailer programs.
07 Sep 08 SOCKET_DEMO.C UPDATED:
The socket_demo.c program has been updated
with new AGILE and IPN notice types.
02 Sep 08 IPN_RAW NOTICES ARE NOW AVAILABLE:
A new type of Notice type has been added to GCN.
It contains Timestamps and Annuli of detections of GRB by instruments
in the IPN constellation of spacecraft.
See the full announcement message.
21 Aug 08 SIMBAD and NED SEARCH ADDED TO AVAILABLE NOTICES:
A new type of Notice type has been added to GCN.
It contains the results of SIMBAD and NED searches
on positions of GRBs reported by other missions.
See the full announcement message.
02 Jul 08 SuperAGILE ADDED TO AVAILABLE NOTICES:
The burst location Notices produced by SuperAGILE are now avaialble.
See the full announcement message.
21 Apr 08 Swift-BAT Slew_Position NOTICES AVAILABLE:
The Harvard group has developed an automated pipeline to analysis the event-event data
collected during the time when swift is slewing to loo for bursts.
Thes burst locations are now avaialble.
See the full announcement message.
03 Apr 08 IMPROVED CATALOG MATCHING:
The ground-catalog source matching has been improved.
The old method scanned the catalog looking for the first source within the matching radius.
This simple method worked fine for several hundred triggers. However, for triggers
306829 and 306830 on IGR J16479-4914, the match returned IGR J16465-4507 (delta=0.28deg),
because it came first in the catalog (before the IGR J16479-4914 source).
Now the search looks for the closest source (not just the first).
This new method works better in the crowded region of the Galactic Center.
15 Jan 08 WRAPPED SUBJECT-LINES IN CIRCULARS NOW HANDLED PROPERLY:
Some user-email-agent programs wrap very long Subject-lines in the email header.
The Circulars processing program could not handle the wrapped portion of these long lines
(that portion was deleted in the out-going circulars being distributed).
The Circulars processing program has been modified to handle these wrapped lines.
(In the prior 7000 circulars, 4 have been received with wrapped lines; with 2 of them
in the last few days (because the Subject-lines for the Transient in NGC 2770 have been
typically longer than the usual Subject-line given the uncertain identity/naming
of this source).) Now all Subject-lines (wrapped or unwrapped) will be processed
in their entirty.
10 Jan 08 TRANSIENT IN CIRCULARS SUBJECT-LINE ACCEPTANCE:
The words "Transient" and "transient" were added to the list of words (one of which)
is required to appear in the Subject-line of every Circular submitted to GCN.
This is response to the X-ray transient in NGC 2770 (which is now called SN2008D).
The full list of words can be found at Circulars.
01 Nov 07 EASIER WAY TO BUILD/CHANGE A SITE'S CONFIGURATION:
A new page has been added that allows the public to (1) build a new site entry,
(b) change an existing entry, or (c) just get a listing in human-readable form
of a current site's configuration. This is located at site page.
28 Oct 07 SKY-ANGLE PROXIMITY TO GALAXIES INDICATION:
For all the position-containing Notices (Swift, INTEGRAL, AGILE, GLAST,
there is now an indication of when a burst position is near/within an NGC galaxy.
See the full announcement message.
30 May 07 NEARBY BRIGHT STAR INDICATION:
For all Swift-BAT/XRT/UVOT Notices, there is now an indication of near-by bright stars.
When there is a star brighter than 6.5 mag closer than 12arcmin to the BAT Position,
a flagbit is set in the socket packets and in the full-format email Notice
there is a COMMENT-line indicating the magnitude and actual distance.
There ~9055 stars in the bright star catalog (from CDS/VizieR); this yields an ~3% chance.
10 Apr 07 Swift-BAT ON-BOARD CATALOG MATCH IDENTITY:
The actual identity of the on-board catalog matches is now reported
in the BAT_POSITION/TRANSIENT Notices. In the past, only a flag bit was set
indicating that there was an on-board match -- now the Catalog ID Number is also reported
(and the source name is reported). The ID Number comes down in the TDRSS packet,
but the source name is determined on the ground by matching to a copy of the on-board catalog.
This information appears in the COMMENTS section of the email Notices.
And in location 25 for the socket site recipients.
28 Feb 07 CELLPHONE SITES CONVERTED TO ALIASES:
All the cellphone & pager sites had their addresses converted to email_aliases
so that they are less likely to get spam from robots gleening from the 'sites.cfg' file.
Since the 'sites.cfg' file is made public on the GCN web site (so people can check their configurations),
email_address-gleening robits can pick up these addresses. And then they are likley
to start receiving spam. Since the alias-defintiion file is NOT made public,
it is no longer possible for this to be a source for spammers to gleen.
28 Oct 06 GCN/COUNTERPART Notice Type:
A new Notice type has been added to the GCN system.
It is the GRB_Counterpart type.
It contains RA,Dec locations of optical,radio,x-/g-ray counterparts
detected by follow-up observers. By having the RA,Dec values in a Notice
(instead of just the Circular), then automated/robotic telecopes can make use of them.
See the full announcement message.
14 Oct 06 SITENAME AMBIGUITY REMOVED:
Over the years some ambiguities in site_names in the sites.cfg file have crept in.
Specifically, sites that have both socket entries and email/pager entries
have developed such that the 'sitename' for one entry was a substring of the other entry.
This caused problems for the end-of-day socket-site report generater. When it scanned
the logfile for the various entries/messages for a given socket-site, it would pick up
entries/messages from the other site that had a superset of the string in the site_name.
I changed about 15 site_names to remove these substring ambiguities.
11 Oct 06 GCN REPORTS ARE NOW SEARCHABLE:
The GCN Reports are now included in the files that are searchable.
Even though they are binary (PDF files), a pre-processing filter was added
to the search hash-table builder, so that all the text in the PDF files
are now scanned and searchable.
01 Oct 06 GCN REPORTS:
The GCN Reports are longer and more-delayed than the Circulars.
Like the Circulars, the Reports contain descriptions of a follow-up teams
measurements and results on a given burst. But the Reports go a step farther
by allowing enough time to pass for a full analysis and full calibration to be done
on the observations. Figures and Tables are allowed. This give observers
that do not have enough observations for regular paper, to publish their full callibrated results
in a citable form, thus allowing other people to incorporate your data into their papers
in a citable form.
See the full announcement message.
16 Aug 06 FASTER GCN MACHINE:
The GCN computer has been replaced by a machine that is 3-5 times faster.
The processing load for Swift GRBs has been increasing over time (last year or so).
The XRt and UVOT instrument teams have been adding new data products (coming down TDRSS)
and new ground processing activities -- all with definite increases in the yield and
quality of the Swft burst response, but this has also impacted GCN's ability
to deliver email-based Notices (the sockets have always stayed fast). The emails have slowed
from 0-4 minutes (a year ago) to 0-10 (sometimes even 15min) recently. Tesing shows
that this delay will be reduced back to the original 0-4 min range.
29 Jun 06 NEW UVOT_POSITION_NACK NOTICE TYPE:
The UVOT Team has created a new notice type: UVOT_POS_NACK.
It will be issued whenever UVOT does NOT detect an afterglow for each BAT burst detection.
See the full announcement message.
29 Jun 06 XRT and UVOT POSITION NOTICES NOW HAVE BAT SOURCE DESIGNATION:
The XRT_POSITION and UVOT_POSITION Notices now have an indication if they are for a source
that was originally identified byt BAT as a non-GRB (ie hard x-ray transient).
See the full announcement message.
31 May 06 ACCIDENTAL DISTRIBUTION OF A TEST UVOT_POSITION NOTICE:
At 19:40 UT 31 May 06, a test of the new UVOT_Position notice was conducted.
This notice was accidentally distributed to the world because of a mistake
in setting up the blocking filters for this test. [I was testing with the UVOT Team
their new ability to generate this new notice type.]
The contents of the distributed notice are totally fake. They do NOT respresent
any burst-related position. Please ignore this notice (retraction emails were sent
to all email/pager sites that originally got the notice and to all socket sites.
29 May 06 SOCKET_DEMO.C UPDATED:
The socket_demo.c program has been updated
with additional and improved comments for the Swift XRT and UVOT notice types.
26 May 06 ADDED SWIFT-UVOT FILTER DEFINITIONS:
The Swift-UVOT SourceList, Image, & Position Notices contain the filter specification
in numerical form. The translation table in english names for thisese filters
has been missing from the sock_pkt_def_doc -- it has now been added.
(The full-format email Notices have always had the filter specification in english (and number).)
26 May 06 FIXED THE EPOCH IN MILAGRO NOTICES:
even though the GCN documentation said the epoch for the RA,Dec in the MILAGRO
socket packets was J2000, what was being distributed was Current epoch.
This was an ~0.09 deg error. (Since the location uncertainty is 0.5 deg, this is a small mistake.)
This also resulted in the same error in the RA,Dec values in the email-based Notices.
14 May 06 FIXED MINUTES IN BAT_LIGHTCURVE EMAIL:
Fixed the "minutes"-units subfield of the TRIGGER_DUR line in the full-format email Notice
for the BAT_LIGHTCURVE Notice. It was off by a factor of 2.5 too big. (The "seconds"-units
subfield was correct.)
01 May 06 FASTER CIRCULARS DELIVERY:
The program that processes the Circulars was changed to increase the speed
of delivery. The old way was processing the email commands to send to each recipient
using a fixed send rate. This rate was choosen to be conservative so as to never
generate a high loadfactor on the GCN machine. But since it had to handle
occasisions of overlapping Circulars distributing, and not to interfer with
the Notices side of the operation, the distribution rate was fixed at a low value.
The new method dynamically checks the loadfactor and spawns a variable rate
of email commands based on the current loadfactor.
The new method is nearly twice as fast as the old method -- 8-9 min instead of 15 min
for 935 Circulars.
26 Apr 06 GCN OPSYS CHANGE OK:
The planned outage to switch to a new operating system on capella
lasted 40 minutes. Both the Notices and Circulars are back on-line.
Capella is now running Scientific LINUX.
25 Apr 06 GCN OUTAGE TOMORROW (26apr06):
The GCN System (both Notices and Circulars) will be off-line tommorrow
Wednesday 26 Apr 2006 from 14:00 to 15:30 UT.
NASA has issued a new set of computer security requirements.
The version of RedHat LINUX that GCN is curently running under
is no longer on the approved list; so I have to upgrade.
GCN Notices & Circulars has already been ported and tested on the new OpSys,
so the transition should go smoothly. 90 min has been allocated
for the switch-over, but it should likely take less time.
(If there are problems, this switch-over is being done in a way
which will allow us to go back to the old OpSys.)
I apologize for the somewhat short notice, but the outage of services
should be small (and I want to get this in with sufficient time before
the weekend).
23 Mar 06 ERRONEOUS BAT LIGHTCURVE FIXED
The GCN system sent out an erroneous lightcurve notice and attachment for GRB 060323.
This was the first example of the "extended lightcurve" for bursts.
The problem comes from the additional prequel data arriving before the original data.
The result was that the values for trigger number, data, time, location had not been updated yet,
so the values left over from the previous burst (060322 in this case) were used and distributed
in this initial copy of the extended lightcurve. This has been fixed.
Also fixed the mistake that alled any of the extended lightcurve notices out to the world.
10 Feb 06 INSTRUMENT MODES & TARGET_NUM ADDED TO THE SWIFT_POINT_DIR NOTICE
The 3 instruments modes (used in burst response data collecting) have been added
to the POINT_DIR notice (both in the socket packets and in the full-format email).
The Target_number and Segment_number have also been added.
08 Feb 06 XRT POSITION UNCERTAINTY IS NOW INTENSITY-DEPENDANT
The way that the position uncertainty for the XRT_Position and _Image was changed
from being a fixed value of 6arcsec to an intensity-dependent based value.
The old value of "6arcsec radius 90% containment" always came out as 5.8arcsec
on the full-format email notices because of the 100microdegree quantization used
within the GCN inter-process communications (ie same as for the socket packets).
The new method uses the GRB_INTEN field in the Image notice to calculate the uncertainty
based on a fit to the previous year's worth of XRT detections.
Example: 45 counts ==> 8.3 arcsec, radius, 90% containment.
07 Feb 06 ADDED FIELDS TO SWIFT_BAT_LC/SM/FOM_OBS/SC_SLEW NOTICES
The Solution_Status, Rate_significance, & Image_Significance fields were added to the
BAT_LightCurve & ScaledMap socket packets and full-format email Notices.
The Image_Significance fields were added to the FOM_Observe and SC_Slew socket packets
and full-format email Notices.
20 Jan 06 AN XRT_IMAGE SLIPPED OUT
During a calibration target pointing today, an XRT_Processed_Image notice
was accidentally distributed. All XRT Notices were supposed to be blocked to world-distribution
during hte 2.8 hour test, but I forgot to activate the XRT_Processed_Image blocking filter.
Receiving this isolated image notice (without having received any of the other standard Swift types)
might have been confusing to people. My apologies. (Please note that during this XRT test
all other function of the Swift mission (including BAT triggering) and GCN distribution
were all enabled and ready for a burst, had there been one.)
18 Jan 06 ADDED SUN_ANGLE FIELD TO FULL-FORMAT EMAIL
To allow follow-up observers to better determine the observability of a given GRB,
the difference in the RA_sun and RA_burst (ie Sun_angle) was added to all the position-containing
full-format email notices (all mission, all types, all sub-types).
The details.
16 Jan 06 ADDED TOO-LABEL TO SWIFT FOM_OBS and SL2SLEW PAGER FORMATS
Fixed the mistake that was preventing the "TOO" field from actually appearing
in the Pager-format Notices for SC2SLEW. And added the same "TOO" field to the FOM_OBS Pager Notice.
05 Jan 06 NO MORE BLOCKING FILTERING FOR COSMIC RAYS IN SWIFT-XRT:
As a result of the GRB 060105 being blocked to world distribution because it was determined
by to be a "cosmic ray" (because of the high gain value in the initial image integration),
the filtering within GCN blocking XRT cosmic ray events was removed. These XRT Position Notices
will still go out with the COMMENT noting they _might_ be cosmic ray events,
but they will no longer be blocked in distribution to the world.
The XRT Instrument Team originially required that these be blocked by GCN. And now they have
requested that this filter be removed.
21 Dec 05 BUG FIXED (that stopped the 051221 SHB):
I have determined what was wrong with the distribution of the 051221 burst.
I totally messed up in integrating a new fucntion into the filtering stage
of GCN. This mistake blocked distribution of Swift-based notices
(both socket and emails) to all non-swift sites (ie the entire world).
I have removed this new function and restarted the GCN program. (It had been
in the running GCN program for about 3 days total -- this was the only Swift
burst during that time.) It will be much better tested before installing
the fixed version.
I appreciate the responses people have sent me about who-got-what-or-not.
Since there were some "world" sites that did get notifications,
I can only assume that you got them through secondary exploders
being operated by people/institutions within the Swift project.
My sincere apologies to everyone for this serious mistake.
09 Dec 05 ADDED LONGITUDE,LATTITUDE TO SWIFT-POINTDIR NOTICES:
This is same as yesterday's addition. But instead of the s/c lon,lat at trigger time,
this is the s/c lon,lat at the time the slew notice is issued (which is about 30-60 sec
before the actual slew.
08 Dec 05 ADDED LONGITUDE,LATTITUDE TO SWIFT-BAT NOTICES:
The longitude,lattitude of the Swift s/c at the time of the BAT trigger
has been added to the bianry socket packets and the full-format email Notices.
This includes the BAT_ALERT, BAT_POS, BAT_POS_NACK, BAT_LIGHTCURVE, BAT_SCALEDMAP,
FOM_OBS, and SC_SLEW notice types.
25 Nov 05 FIXED FLUX MISTAKE IN 2 SHORT PAGER FORMATS:
Two of the 4 Short-Pager formats had the Flux in the XRT_POSITION Notices still in the old arbiray units
instead of the CGS units (see 14nov05). This has been fixed.
20 Nov 05 CHANGED THE TIME FOR SUN/MOON CALCULATIONS:
In the beginning days when all burst notices were created at the time of the burst,
the use of the 'burst time' for calculation of the Sun/Moon RA,Dec position and
the Moon Illumination values was fine. But now there is a small fraction of the Notices
which are created with some time day (minutes to hours). (HETE GndAna, Retractions, and
INTEGRAL Test Notices are these delayed notices.) As such, the Moon will have moved
between the time of the burst and the time of the notice. The GCN programs have been changed
to use the notice time (ie the current time) when calculating the Sun/Moon positions
and illumination values. The details.
19 Nov 05 FIXED MISTAKE IN THE SOCKET_CONNECTION_REPORT:
Since Feb 05 (essentially the beginning of the distribution of the Swift Notice types),
there has been a mistake in the counting of those Swift types. Only part of the types
were counted and reported. So you can not rely on those totals from those reports,
nor the contents of the 1st & 2nd histograms. (The error_mesg counting has been correct.)
This mistake has been fixed.
Also fixed the mistake of double counting the out_of_nnn values.
A 4th, 5th, & 6th histogram were added (of Non-Pos_Swift, all_types_your_site & all_types_all_sites).
14 Nov 05 XRT_POSITIONS Notices Go To Real CGS Flux Units:
This announcement describes the details of the switch-over
of the GRB_INTEN field in the XRT_POSITION Notice from 'arbitrary' units to 'erg/cm2/sec' units.
05 Nov 05 SWIFT STAR TRACKER COMMENT BUG FIX in the Notices:
A bug was found that caused the Swift-BAT Position email Notices to be created
with the "StarTracker Not Locked" comment to be included. The logic of that flag bit
was inverted. (Eg, the 051105 short burst notice this morning went out with an incorrect
statement about the StarTracker and "possible bogus detection"); ditto for the Transient Notice at 15:49 UT.
The flag bit is also inthe socket packet format.
The GCN program was restarted (with the new/fixed code) at 18:28 UT today.
03 Nov 05 LABEL CHANGE on XRT_POS, XRT_IMAGE and UVOT_POS Notices:
This announcement describes the label changes on the DATE and TIME fields
in these 3 notice types.
03 Nov 05 BUG FIX in XRT_IMAGE Pager-format Notices:
The "SIGMA=" field in the XRT_IMAGE pager notices was printing the wrong value.
This field has been changed to "ERR=" and the correct value.
Plus the "I=" (for intensity) field was added (with units of "[cnt]").
23 Oct 05 STAR TRACKER LOSS-OF-LOCK in Swift NOTICES:
A flag-bit was added to the "solution status" field (aka TRIGGER_ID)
for the SWIFT_BAT_POS,_LC,FOM_OBS,SC_SLEW socket packets. This flag-bit
also translates into a COMMENT line in the full-email format Notices.
17 Oct 05 RETRACTION LABEL ON INTEGRAL NOTICES:
The full-email format and the pager/cell formats of the INTEGRAL notices
did not have the "_RETRACTION" modifier attached (when INTEGRAL IBAS issued
a retraction message). Now these notices do have this modifier.
Also added the description of this flag bit to the sock_pkt_def_doc.
25 Sep 05 SCALING CHANGE IN UVOT SOURCE LIST:
The scaling was changed for the "bkg_mean" quantity in the socket packets
from 1 to 1000. This is needed since it turns our that this item is often less than 1.
05 Sep 05 BUG FIX FOR INTEGRAL TEST NOTICES:
There was a bug in the filtering code that was blocking INTEGRAL Test Notices
to recipients. There was an interaction between the Time_Delay filter and the Notice dates.
See the details.
24 Aug 05 EMAIL RETRY ATTEMPTS ADJUSTED AGAIN:
Two days ago the sendmail retry-to-send time interval was changed from 60 to 10 minutes.
Further study has uncovered the rare possibiltiy that for some attempts
10 min may be a tad too short, resulting in an overlapping retries.
So, the retry time has now been lengthened to 15 min.
This is not a big deal, because this retry function applies to only those email sites
that did not respond to the sendmail demon on the first attempt (which is rare).
22 Aug 05 EMAIL RETRY ATTEMPTS SHORTENED:
The email that GCN generates is sent recipients (full format, cell, pager, etc).
If the initial attempt fails (a) to establish the 'sendmail' transfer link
to the host machine of the recipient and/or (b) determine if the recipient
is a current/valid account, then it keeps trying for up to 2 days.
This retry duration was 5 days, but it has now been shortened to 2 days.
The reason for this decrease is that given (a) the volume of email distributed,
(b) the number of recipients, and (c) the fraction of unreachable_machines/defunct_accounts
the number of email "jobs" stacked up in the sendmail outgoing queue is getting large.
This large number was consuming disk resources and cpu time (in the many repeated attempts).
The main reason for the growth in the cpu-usage was due to also changing the retry time
from every hour to 10 minutes (to speed up delivery to the momentarily-unavailable sites).
(Another way to reduce the usage/load was to remove some of the entries in the distribution list
that repeatedly fail and have so for more than 6 months (ie they are truely "off net").
I do this pruning about once a year, but it is very time consuming, and so I need
a way wich is more global in its effect/coverage.)
20 Aug 05 SMOOTHER START-UP FOR SOCKET SITES:
When the GCN program is (re)started, it immediately tries to connect to all the socket sites
in the sites.cfg file. And if the site connected on that first attempt, then it would also
immediately get sent an Imalive pkt, a Test(type=2) pkt, a HETE Test pkt, and a Swift Test pkt.
This onslought of pkts might have been a problem for some sites. Now the GCN program
delays the sending the initial Test pkts (Imalive still comes immediately), and there is a 5-sec
separation between them. This should smooth out the 'load'. (And it makes things easier
at the GCN end when all those echoed-back pkts start coming in the from the connected sights.
16 Aug 05 STAGGERED SOCKET-SITE RECONNECT ATTEMPTS:
Out of the 85+ socket sites defined in the "sites.cfg" file, at any given instant
there are ~45 routinely connected and ~40 not connected. The gcn program periodically tries
to connect to these non-connected sites. The program used to have a progressive series of connect times:
T+0 minutes (ie at the gcn program start-up time), T+4 min, T+8 min, T+16, T+32, T+64, T+120, 240, 360, 480, 600, ...
This series meant that all the non-connected sites were attempted in total phase-locked occurances.
So when one of the times came up, all ~40 sites were attempted (one right after the other).
Since it takes a maximum of 3.0 sec for a socket connection attempt to time out,
this meant there could be up to a 3*20=60 sec window of time when GCN was effectively incapable
of distributing Notices to the rest of the list. This phase-locked connection algorithm
was simple to implement and has worked well for ~14 years. But now in the Swift-era
a 60-sec deadtime window is too large. The breaking of the phase-locking was accomplished
by adding a linear term to the geometric term to the calculation of the connect_time series.
The linear term is based on the socket site's location in the list in the sites.cfg file.
The time series of reconnect attempts is like the above but with this linear term ranging
from ~10 sec to ~6 minutes.
10 Aug 05 SPEED IMPROVEMENT IN SOCKET DISTRIBUTION:
From the original days of GCN (ie BACODINE back then), the socket-site entries
have been put first in the list of recipients (ie the "sites.cfg" file),
and the email-based recipients have come later in the file. This is done this way,
because the GCN program loops through this file (top down) when distributing
any type of Notice. Since email-based recipients are human and socket-site recipinets
are robotic telescopes, it makes sence to get the information to the robotic telescopes
faster because they can respond faster than humans. But over the years, a mixing
of these two groups of entries has occurred. This happened because some "sites"
wanted both socket and email distribution methods. And I put the two entries
in the file next to each other so I, as adminstrator, could keep track of "groups"
of people/operations/etc. But this intermixing has grown too large, and now
with the advent of the "comments" column in the sites.cfg file, I can keep track
of those multiple-entry groups without needing juxtoposition in the file.
So I have made a complete/clean separation of the socket sites all up front
in the file list, and the email-based sites coming second in the file.
The speed enhancement is about 0.15 sec for the last of the socket entries
(this was 0.15 sec on top of about 1.0 sec). Granted, this is not much in the total scheme
of these burst follow-up activities, but it is still worth doing as there is
no longer any down-side to doing this change.
09 Aug 05 SPEED IMPROVEMENT IN EMAIL DISTRIBUTION:
The way the many email-based Notices (full email, pagers, cell's, etc) were being distributed
to the GCN client list was changed. The old way was fine when the number of recipients
(ie number of email cmds) was small (ie 50-100), but now in the Swift era
when more people have enrolled into GCN, the number has grown to ~250
for the most popular Notice types. For 250 email commands, the total execution time was 4 minutes.
The new method has cut that time by a factor of 2.5 (ie 1.6 min from the first recipient
to the last recipient).
24 Jul 05 MISSING XRT_POSITION NOTICES (since about 2 weeks ago): Some people were not getting their copies of the XRT_POSITION Notice. This was a complicated problem that has to do with subtle interactions between the (a) the 'sh' shell, (b) the internet, and (c) the details of the configurations of the entries in the sites.cfg file. It took 3 days to find (and would have taken longer if I had not had access to a machine outside GSFC -- A special thankyou to Carlo Graziani for allowing me to use his machine at U.Chicago).
23 Jul 05 MISSING PACKETS IN MULTI-PKT TDRSS MESSAGES (since Day 1): The BAT_SCALEDMAP, UVOT_DARKBUST, and UVOT_FINDCHART TDRSS messages are multi-packet messages (ie 36, ~20, 1-3, respectively). The problem was that sometimes (about 9in10, 1in2, and 3in4, resp) there would be a missing pkt. This was not a big problem for the SCALEDMAPs because the coded-aperture process is forgiving of "holes" in the data. But is was a significant problem for the DB and FC because they are compressed and once the encoding sequence it broken, it can not be resumed later on in the sequence when the packet sequence resumes again.
12 Jul 05 Improved version of socket_demo.c. Improved comments for the many Swift-based TRIG_ID and MISC flag bits.
05 Jul 05 The PDF format as addd to the list of formats available for the Swift-based lightcurve, images, spectra attachments (GIF, JPEG, PDF, Postscript, text, FITS).
13 Jun 05 Fixed a mistake that the Error field in the SubjectHHMMSS_Pager format for the Swift-BAT Notice type was always zero.
10 Apr 05 The distribution speed for Circulars has been increased.
Over the course of time the distribution time has grown to be ~35 minutes to send the emails
to the 860+ recipients. I have restructured the program to make smarter use of the cpu load-factor and
n_process_per_user such that the time is cut in half. It now takes 17 minutes for all 860+ recipients.
(Work will continue, when time permits, to implement another speed-up algorthm which should reduce
the time another factor of ~2.)
01 Apr 05 Fixed the mistake that was sending two copies of the Swift-BAT_Position Notice.
24 Mar 05 GCN/Swift-PointDir Notice now available. See the announcement for more details.
26 Feb 05 GCN/Swift-XRT goes real-time distribution (no human delays) announcement. See the announcement for more details.
26 Feb 05 The socket_demo.c program has been updated with better/additional parsing of the flag bits and the comments.
14 Feb 05 GCN/Swift-BAT goes real-time distribution (no human delays) announcement. See the announcement for more details.
18 Jan 05 Fixed a bug which prevents about a third of yesterday's Swift-BAT_Pos email Notices from being distributed. It was a mistake which coupled attachments and mail-vs-mutt.
21 Dec 04 Call_for_GCN/Swift_Sign-up announcement. See the announcement for more details.
16 Dec 04 Swift Pointing Direction is on-line.
During the initial Launch-n-Early-Orbit Checkout/Verification phase,
Swift will be doing some planned pointing observations (no new burst automated targets yet).
These pre-planned pointing are listed in the Swift PPST.
This table will be updated (manually at first) periodically. (Later, it will be auto-updated
as soon as the PPST files become avaialble from the Swift MOC. And eventully these pointing
changes will be fully automated to the point where the GCN/Swift_Point_Dir Notices
will be "pushed" to those sites/recipients that want to receive them.)
16 Dec 04 Fixed a bug in the auto-updating demon for the GCN Circulars archive page that was triggering on the Swift-XRT-* messages. The old demon filter rule was set up to trigger off the "XRT" string in the Subject-line of the Circular (just like the "GRB", "SGR", etc strings). But now with the Swift-XRT GCN Notices, this rule was incorrectly picking up these as well. The filter rule was changed to be more slective on Circulars and not pick up on the Swift-XRT Notices.
06 Nov 04 Announcement about Swift being integrated into the GCN system. Also information about the formats and contents. See the announcement for more details.
10 Sep 04 The machine used to process the GCN Circulars (both submissions and distributions) was changed from 'lheawww' to 'capella'. People are strongly encouraged to send their circular submissions to "gcncirc@capella.gsfc.nasa.gov". The old "gcncirc@lheawww.gsfc.nasa.gov" will still work (there is a forwarding file) and it will be maintained indefintely, but since nothing is really stable for infinity in computers systems, you are strongly encouraged to change your aliases/etc to the new capella-based address. See the 18sep04 announcement for more details.
26 Jun 04 (a) Improved the socket_site handling code
to allow for the scenario where the sites echo-back the packet rotated
to the byte-order for their own platform. These byte-rotated packets
used to cause an error_message in the returned-packet checking code.
And this error ended up in the daily Socket connection Report. Since there 1440
of these messages every day for each of those sites, their Reports were cluttered up
with somewhat useliess error messages, and then the 1 true error message they needed
to see was lost.
(b) Added the NYTT declination-specific custom filter.
30 Jul 03 Add the system status link to the beginning of the main page. This link replaces the actual placement of the system status messages at the beginning of the main page.
20 Mar 03 Fixed the same typo on the new system that caused the 2.9armin error on the old system in the INTEGRAL coords. In the hectic switchover (neccessitated by the IT security folks here at Goddard after the hacker incident), the fix done on the old system on 01 Mar 03 did not get moved to the new system -- now it has.
11 Mar 03 The GCN system switch-over to a new computer has been completed. GCN is up an running. See the details.
01 Mar 03 The INTEGRAL GRBs table now updates in real-time. Fixed a typo in the INTEGRAL coordinates conversion that produced a 2.9armin error. Finished the INTEGRAL web page herein.
21 Feb 03 The GCN system will switch over to a new computer
on 11 Mar 03. There are no changes for any e-mail-based site. And only the socket sites that are behind a firewall will have to make a change: the new machine's
IP Number will have to be added to their firewall pass-through list.
Details of the GCN Machine Switch-over
14 Feb 03 The GCN system was off-line for 23 hours. The details are given in here.
13 Dec 02 The GCN system was off-line for 16 hours (00:00 to 15:58 UT Friday, 13 Dec 02). The reasons for this outage are as yet undetermined.
19 Oct 02 GCN was off-line (both Notices & Circulars) due to a GSFC-wide network system upgrade. It was off-line from 14:11 to 15:07 UT, and sluggish from 15:07 to 15:40 UT.
12 Oct 02 The Circulars processing demon was modified to send to the successful submittor an email that comes immediately. Prior to this change, the successfull submitor had to wait until the Circular had been sent to the entire list to find out if he/she was successful. That old notification is still sent, but now you get a message before all the Circulars are distributed. This greatly shortens the time you wonder if you were successful. (Of course the problems-with-submission notification still comes immediately.)
03 Sep 02 The INTEGRAL burst location Notices have been added to the GCN system. See INTEGRAL details and the Circ #1529 announcment for the full details.
20 Jun 02 The list of strings that are allowed/required for GCN Circular submission has been increased to include "AXP" (anomolous x-ray pulsars).
15 Apr 02 The IPN is back in business -- there are 3 spacecraft. The GCN/IPN Position Notices are being distributed (in all forms: socket, email, & pager). For the details. And for the details of the GCN/IPN_Position Notices.
08 Apr 02 The HETE Notices (emails and socket packets) now have several additions: the WXM Image X&Y; Signal/Noise, the WXM Lightcurve X&Y; Signal/Noise, the S/C Longitude, and the Ecliptic Lon,Lat of the burst location. See the details.
11 Nov 01
The list of strings that are allowed/required
for GCN Circular submission
has been increased to include new transient types and mission-based strings.
See the details.
And a few days after that, the "HST", "SDSS", and "XRB" strings were added
to the list.
16 Aug 01 The KONUS LightCurves page now has Trigger Number and Trigger Type information added to the table. These additional fields for each trigger as supplied (by human intervention) about 2 weeks after the event.
06 Aug 01 The HETE Ops team is no longer distributing trigger Notices which are due to x-ray burst sources (since several days ago). This explains the dramatic decrease in the number of GCN/HETE Notices. However, if HETE triggers on a GRB, it will still be distributed to GCN (and therefore to the world).
30 Jul 01 Status of HETE and the various GCN/HETE Notices. From the HETE perspective: Circular #1084. From the GCN perspective: Circular #1085.
28 Jul 01 The filtering of HETE_ALERT Notices was changed to allow sites to get them even though they have no position. Previously, they were blocked because they had no position (at least most of the times they had not position -- about 10% they do), but now they will be distributed to those sites requesting the HETE_ALERTs even when they have no position information in them.
20 Jul 01 The GCN systems back to normal. There was another 36-to-48 hour "data storm" inside Goddard's Build 23 again (just like 16-17 July), which is now gone. But I also changed the socket-connection code to reduce the slow-down impact of these storms.
18 Jul 01 The GCN systems back to normal. The Internet connectivity and responsiveness problems reported yesterday have disappeared. The connections have been back to normal for over 18 hours which I take to mean that whatever "data storm" was going on during late 16 to mid 17 Jul 2001 has disappeared.
17 Jul 01 The GCN system has been experiencing network connectivity and throughput problems (breaks, slow-downs and blockages) since early Monday morning (16 Jul 01). Investigations are underway -- I will keep you posted. This affects the socket-connection sites the most.
11 Jul 01 Fixed the bug in the GCN s/w that caused the "no position available in this notice" when there was, in fact, a position available in the HETE Notice.
28 Jun 01 The GCN system was off the air for 27.7 hours; the details.
20 Jun 01 See here for the announcement of the label changes for the HETE email/pager Notices.
16 Jun 01 See here for an explanation of the difficulties with processing HETE triggers 1554 and 1555.
10 Jun 01 The HETE Triggers Information web page is now fully automated. Like all the other Mission-Instrument GRB web pages, this is also updated in realtime (within a few seconds) after the GCN Notices are distributed.
31 May 01 Full automatic operations with HETE have started. The process is now completely automated from the on-board calculation of the burst position, to the transmission to the ground and GCN and on to those sites electing the receive the GCN/HETE Notices. See the full announcement. Also see the HETE web page in this GCN web site.
12 Jan 01 Fixed a mistake in filtering code that caused some sights not to be sent some GCN/SAX-WFC Notices during the Jun 2000 to Jan 12, 2001 six month interval. See the full announcement.
11 Jan 01 Fixed a mistake in code that caused the wrong TDJ given in burst notices that occurred in the year 2001. They were off by 1 day. It had to do with the year after the leap_year mod_400 testing. In practice, this affected only one IPN_POSITION Notice, as there were no other burst-generated Notices in the first 11 days of this year. (It did affect the Test Notices and the Imalive packets.)
16 Dec 00 Changed the maximum socket_reconnect retry time interval from 30 minutes to 60 minutes.
18 Sep 00 The IPN_POSITION burst location Notices have been added to the GCN system. See IPN details and the announcment for the full details.
26 May 00 CGRO Instruments off (loss of BATSE and COMPTEL positions):
GCN announcement
NASA HQ press release.
25 Jan 00 Added links to PostScript versions of the various published papers on the GCN system. See the references page.
09 Jan 00 There are two new filter available to sites to determine which Notices they get sent. They are based on the size of the error box (location uncertainty) and on the amount of time which has elapsed between the start of the GRB and the time of Notice distribution. These two new filters are applied independent of any other filtering criteria already available. See the full announcement.
03 Jan 00 The GCN system is back on the air after the Y2K and Millennium Madness outages. See the full announcement.
23-26 Dec 99 and 30-03 Jan 00 Huntsville RBR LOCBURST notices will be off-line, and there is some Y2K issues for 30 Dec to 03 Jan. See the full announcement.
18 Dec 99 The bug that caused a delay in delivery (minutes to an hour) of GCN 471 has been fixed. It was due to the inclusion of single (or double) quotes in the Subject-line of the circular submission. The new program handles single and double quotes just fine, so submitters are free to use them.
09 Dec 99 CGRO lost a gyro and had a pointing table problem which caused the loss of 3.2 days of telemetry. The details of the outage are explained in this announcement.
04 Dec 99 The BeppoSAX WFC & NFI burst location Notices have been fully automated. Since late 1997 they have been only semi-automated (with a human-in-the-loop at the GCN end), but now the BeppoSAX and GCN teams have completed the process to be fully automated end-to-end. See announcement for the details. See BeppoSAX details and the BeppoSAX Locations Table.
13 Nov 99 The graphical form of the BATSE lightcurves have been restored to the GCN system. See announcement for the details. These lightcurves are available within the BATSE Triggers web page.
11 Nov 99 A series of network problems has caused a series of losses of service for the GCN system. The detail of these outages is explained.
07 Nov 99 The GCN/NEAR Notice light curves table is available on the GCN web site. This table (like the BATSE, XTE, KONUS, etc table pages) lists all the NEAR-XGRS detections and provides the light curves in Postscript, JPEG, and numerical text formats. There are more details in the announcement notice.
05 Nov 99 A failed internet connection to one of the GCN computers (the Channel Controller) failed and replaced with a mini-hub. Total downtime was 20 hours. See the notice.
29 Oct 99 Added the ERROR_2m filter. It allows sites to receive burst location Notices that have an location uncertainty less than 2 arcmin (radius). This filter applies to any source-type that site has enabled.
10 Oct 99 Created the QUEST filter. It is another example of a site-custom filter.
09 May 99 GRO telemetry lost for 7.5 hours on 09 May 99. For the details see GRO5_09.TXT.
08 May 99 The delay in distributing the BATSE-Final socket packets was decreased from ~120 sec to ~45 sec (the best case possible is 35 sec). This was done by splitting the socket-site processing from the email-site processing within the GCN program. See the full announcement.
20 Mar 99 The new SubjHHMM-only distribution method was added to the existing list of distribution methods (Internet socket, e-mail, pagers, and phones). This new method puts the RA,Dec,Time,Intensity plus the source of the location (BATSE, RXTE, SAX, etc) in the Subject-line of an e-mail that is sent to your pager or cell-phone (which is only capable of displaying the (typically) the From-line and the Subject-line of incoming e-mail). There is a full description of the Subject_HHMM-only Distribution Method. Note that the RA,Dec are in a different-than-normal format: RA=+HH:MM:SS Dec=+DD:MM:SS.
14 Mar 99 Created the EROS filter. It is another example of a site-specific-custom filter.
27 Feb 99 Added the ERROR_1d filter. It allows sites to receive burst location Notices that have an location uncertainty less than 1 degree (radius). This filter applies to any source-type that site has enabled.
23 Feb 99 The SEU anomaly with BATSE has been fixed (GCN Circ 265). Normal operations resumes.
22 Feb 99 The BATSE instrument is experiencing anomalous behaviour in its flight swatch (which produces MAXBC Notices with bogus C1 and BC rates and, therefore, bogus locations). This is described in GCN Circ 264. Please do not use these notices for any follow-up activities.
29 Jan 99 The RXTE-ASM instrument has set up an automated data processing scheme to search for and localize x-ray transients within its FOV. Based on archival data searchers, there will be about 8 of these transients (Novae, GRBs, and other types) per year. They will be localized to within 6 arcmin. They will be distributed via the GCN as a new Notice type: ASM_XTRANS. GCN sites are encouraged to sign up for this new Notice type.
27 Jan 99 CGRO-BATSE changed its trigger criteria back to the standard channels 2+3 (50-300 keV). The first trigger using the new trigger channels was #7356, an accident caused by the flight software change. [Previously, BATSE had been triggering on channels 3+4 (nominal energy range greater than 100 keV).]
07 Jan 99 There was a problem in the "date" field of the BATSE Notices for Triggers 7293-7300 (01-04 Jan 99). This was due to the fact that the "front_end" telemetry processors in PACOR do not know the year. (They know the the month, day, hour, minute, sec, and even the microsec, but they do not know what year it is.) To compensate, I have the year hardwired into the program. I also have code to handle the transition at mindnight 31 Dec, but because of another (simultaneous) problem with the clock generator hardware, the transition software got confused and generated "odd" dates for a few notices. It took me a fewe days to notice this new problem and fix the software (both the year-patch code and the transition code (so it will work even better next 31 Dec).
29 Oct 98 The GCN/KONUS Notice light curves table is available on the GCN web site. This table (like the BATSE, XTE, etc table pages) lists all the Wind-KONUS detections and provides the light curves in Postscript, GIF, and JPEG formats.
21-26 Aug 98 There was an anomoly in the GRO s/c pointing which caused a loss of tracking on the desired target. Several days were required to figure out the fundamental problem (most likely due to a change in the GSE software). It should be noted that there was never any problem with the spacecraft itself (hardware or onboard software) so the life expectancy of GRO is still good. Normal observing and data flow from the instruments resumed Wednesday afternnon (26 Aug 98).
08 Aug 98 The GCN system was off-line from 01:00 until 15:30 UT 08 Aug due to a Goddard network problem.
07 Jul 98 Changed the plotting options that make the postscript and GIF versions of the BATSE light curves so that zero is surpressed on the y-axis. This surpression of the ambient background count rate in BATSE gives emphasis to the burst light curve.
14 Jun 98 Added GIF-ed version format to the available formats of the light curves for BATSE triggers (that have enough counts to calculate a location and type).
02 Jun 98 Added "SGR" to the list of acceptable keyword strings in the Subject-line of the GCN Circulars processing (so now the list is "GRB", "grb", and "SGR").
31 May 98 The triggering criteria on BATSE was changed from the usual channels 3+4 to channels 1+2. See GCN Circular #93 for the details.
19 May 98 The triggering criteria on BATSE was changed from the usual channels 2+3 to channels 3+4. See GCN Circular #76 for the details.
27 Apr 98 The electrical power work for building 23 has been completed. The GCN computers were rebooted, the internet network connections were re-established, and the CGRO telemetry processing resumed. This is the last of the scheduled outages. Not counting these two planned outages in the last month, the GCN system has a >98% livetime record.
24 Apr 98 Two of the three parts GCN system will be off-line (unavailable) between Friday 19:00 UT 24 Apr 98 and Monday ~11:00 UT 27 Apr 98. This is the second of a two-part process to replace the main power transformer substation for building 23. Building 23 processes the telemetry for all the near-earth spacecraft (i.e. GRO, RXTE). The GCN Circulars operation (part 3) is housed in a different building, so it will continue to operate during this interval.
10 Apr 98 The GCN Circulars system was automated. The announcement was made via Circular 54. People can now submit and receive GCN Circulars via the automated system. However, the manual method will be maintained for those individuals who have not yet given the extra information for them to be transferred from the manual list to the automated list.
23 Mar 98 The electrical power work for building 23 has been completed. The GCN computers were rebooted, the internet network connections were re-established, and the CGRO telemetry processing resumed. So the full set of GCN capabilities have been resumed (as of ~12:35 UT).
20 Mar 98 Two of the three parts GCN system will be off-line (unavailable) between Friday 17:00 UT 20 Mar 98 and Monday ~11:00 UT 23 Mar 98. The main power substation for building 23 here at GSFC is being replaced due to excessive age. Building 23 processes the telemetry for all the near-earth spacecraft (i.e. GRO, RXTE). The GCN Circulars operation (part 3) is housed in a different bulding, so it will continue to operate during this interval.
22 Feb 98 The GCN Circulars announcement and call for sign-up is made. This is the culmination of the preliminary announcement, the 4th GRB Workshop special interest session discussions, and the following months discussions. A call for people to sign up to the submission (validation) and reception list was made. (IN a couples weeks the transition from the current manual mode of collecting & distributing the Circulars ito the automatic mode will be complete.)
25 Jan 98 The new version of te GCN web site was activated (actually much of these work as been incrmentally added over the last 5 months). The organization of the main page was changed and many of the top/mid/detailed level description pages were rewritten to reflect the evolutionary changes from the old BACODINE system to the new GCN system. Many format and procedure pages were added. The total page count has just about doubled from the original BACODINE version of the web site.
01 Jan 98 I forgot to activate the year-rollover code which patches the lack-of-year-knowledge for the GRO telemtry processors. Hense, the GRB_DATEs for BATSE triggers 6553 and 6554 had the wrong DOY and YY/MM/DD. This has been fixed, and all will be well for another 365 days.
18 Dec 97 The LOBURST notices will not be available between 26-29 Dec because of the holidays and Huntsville personnel scheduling. All other GCN notice types will continue to be generated and distributed.
Scott Barthelmy will be unavailable between 22-29 Dec. All questions and requests for action should be directed to Paul Butterworth during this period. (It is a good idea to CC Paul on all your GCN-related e-mails at all times anyway.)
17 Dec 97 A "shortcut" links section for the "recent" burst was added to the main web page. It has all the BATSE location & light curve information, plus IPN location, plus GCN Circulars optical/radio/xray follow-up information. This provides the GRB researcher with a quicker way to access all the GCN information about the hot-burst-of-the-week.
24 Nov 97 The GCN/COMPTEL GRB Notice type is available for distribution to GCN sites. For bright bursts that are in the FOV of the CGRO-COMPTEL instrument a location is calculated by the UNH COMPTEL team from the telemetry data which is captured by the GCN system and transferred to UNH (about 5 GRBs/year). This UNH location is sent back to the GCN system for distribution to the GCN sites.
04 Oct 97 The BATSE-Final GRB Coordinates Notice type is available for distribution to GCN sites. The Final Notice type is similar to the BATSE-Original type except that it is based on an integration of the light curve for up to 32 seconds after the initial BATSE trigger. The peak intensity and fluence during that interval are also reported. These two items allow the user to access the "importance" of a burst when deciding about making follow-up observations.
11 Sep 97 The BATSE GRB Information Table web page is created. It contains RA,Dec, Data,Time, Intensity,Fluence, Light Curve information about all BATSE triggers (GRBs and non-GRB triggers) processed by the GCN system. It is updated in real-time.
05 Jul 97 New copies of the sites.cfg file can be loaded into the GCN program dynamically (while the program is running) and it can be done remotely. This means I don't have to go from building 2 to 23 to restart the program to get the newly updated config file active.
02 Jul 97 The new Subject-only distribution method was added to the existing list of distribution methods (Internet socket, e-mail, pagers, and phones). This new method puts the RA,Dec position plus the source of the location (BATSE, RXTE, etc) in the Subject-line of an e-mail that is sent to your pager or cell-phone which is only capable of displaying the (typically) the From-line and the Subject-line of incoming e-mail. There is a full description of the Subject-only Distribution Method.
23 May 97 Two new location notice types were added to the system. They are the RXTE-PCA GRB location notices and the ALEXIS ultra-soft x-ray transient locations. The RXTE-PCA locations are derived for bright GRBs using the LOCBURST Notice locations and also passing through the RXTE scheduling process. There is a full description of the RXTE-PCA Notices.
The ALEXIS notices are a variation on the original GRB-only theme of BACODINE. The ALEXIS spacecraft has 6 telescopes in 3 energy bands (66, 77, 93 eV) and it locates 0-3 unidentified transients every week. There is a full description of the ALEXIS Notices.
BACODINE changes its name to GCN (GRB Coordinates Network) to reflect the new broader scope of the types of GRB (and ultra-soft x-ray) transient location information now available.
02 May 97 A new GRB Location Notice type was added to the system. These LOCBURST locations are derived for bright GRBs using the Huntsville Locburst algorithm/procedure. It is a semi-automated process yielding 2-3deg error radii locations in 15-25 minutes time delay. There is a full description of the LOCBURST Notices.
25 Feb 97 The BATSE on-board flight software triggering criteria were changed back to the normal mode (optimized for classical GRBs); that is to say: channels 2+3 (50-300 keV) at 5.5 sigma in any of the three times scales (64, 256, 1024 msec). For a list of all the BATSE Trigger Criteria changes.
12 Feb 97 Added a discussion on the duration and frequency of gaps in the telemetry from the GRO s/c.
01 Jan 97 Added GRO SSC Status Reports archive page. The archive runs from the latest Status Report back to Mar 93.
05 Nov 96 The on-board trigger criteria for BATSE were changed from the normal 50-300 keV band with a two-detector 5.5-sigma threshold to 20-100 keV (still 5.5 sigma for the 64 & 256 msec timescales and 7.0 sigma for the 1024 msec timescale, & all three scales require two detectors).
14 Oct 96 Added BACO/IPN Notice archive page. Currently, all the BACODINE/IPN e-mail notices that have been sent are archived here. In the future this page will be automatically updated as each new Notice is sent.
24 Aug 96 A new distribution method was added to the list. It is called the "Short Pager" method, because it is a shortened message form of the original alpha-numeric pager message. The original message contained the RA,Dec,Intensity,Time; the short form has only the RA & Dec. This works for those people/sites that have pagers with a limited display size.
31 Jul 96 The socket_demo.c program now automatically alerts the operator (via e-mail) when the imalive packets have stopped (and when they resume).
20 Jul 96 Added the capability that a special Test Coordinates packet is sent to the Bradford Telescope socket site at 01:00:01 UT everyday. (This is different than the normal Test Coordinates packet sending capability to any site that elects to receive them.)
26 Jun 96 An intermittantly bad multi-ether-port mux box was replaced. This bad mux was causing slow ftp transfers of COMPTEL data and may have caused about 1-in-10,000 Imalive packets to get garbled.
25 Jun 96 The on-board trigger criteria for BATSE was changed from the normal 50-300keV band with a two-detector 5.5-sigma threshold to 100-infinity keV and 4.5 sigma. This change was made to greatly enhance the sensitivities to Terrestial Gamma Flashes (TGF) during a multi-wavelength and multi-investigator observation campaign on TGFs. The sensitivity to classical GRBs is only decreased by about 10%.
25 Jan 96 The BACODINE software was changed so that it can now retrigger and calculate locations during a BATSE Trigger Readout. BATSE could always retrigger during this readout period if the second trigger was brighter than the first trigger, but BACODINE was dead during this readout interval. Now BACODINE can retrigger as well. This is particularly useful during all these triggers from GRO1744-28.
Earlier upgrades are listed in the History & Evolution of GCN.