LoadRunner
Release Notes, by Version

LoadRunner v 1.2b
October 13, 2011

• Bug fix:
  • Mac version no longer installed or ran, as of Mac OSX Lion. No more PowerPC support.

LoadRunner v 1.2a
April 22, 2009

• Bug fix:
  • Full year = false option was broken since version 0.7e (I think).

LoadRunner v 1.2
January 15, 2008

• Enhancement:
  • Unit conversion. On the "ElemNum" list on the LoadRunner screen, you can add a unit conversion to one or more USGS element codes. This is useful for munging data as "same substance" when their units are different. For example, if you wanted to combine 00910 CaCO₃ - which is 40% Ca by molecular weight - with 00915 Ca, you could enter on the ElemNums:
    00910x0.40,00915

    This could also be used to munge measurements in micrograms and milligrams. But - convert up, not down. The calibration data file uses only 3 digits past the decimal place. So it's better to use "x1000.0" on the quantity in micrograms, than use a conversion of "x0.001" on the milligrams.

  • For clarity, added "(or g/day if input was in micrograms)" etc. type comments to the output.

LoadRunner v 1.1b
December 6, 2007

• Enhancement:
  • New option to tell LoadRunner to pass non-detects through for LOADEST to handle. This is important for constituents with a large number of non-detects (pesticides, for example). However, it's been known to hang LOADEST (and thus LoadRunner) on bad data. See the LoadRunner Manual regarding the "< detect" option for further discussion.

LoadRunner v 1.1a
December 3, 2007

• Enhancement:
  • LoadRunner now runs on Intel Mac's, PowerPC Mac's, and Windows. Has been tested on Mac OS X 10.5 (Leopard) on Intel Macs.
  • Thanks to Robert Max Holmes of Woods Hole for supporting the Mac ports of LoadRunner/LOADEST.

LoadRunner v 1.1
November 15, 2007

• Enhancement:
  • LoadRunner runs on Intel Mac's as well as Windows (tested on OS X 10.4 Tiger; not tested on 10.5 Leopard yet). Please note that it does not run yet on PowerPC Macs (G4 processor, etc.) I don't have a version of LOADEST that works on those yet.
  • Thanks to Robert Max Holmes of Woods Hole for supporting this LoadRunner/LOADEST port.
  • A couple minor side effects of this:
    • Underscore characters in site name comments are now replaced by blanks.
    • Caveats like "<" or "F" are no longer written into LOADEST calib.inp files. These comments are written to LoadRunner's output files, though.

LoadRunner v 1.0a
October 9, 2007

• Enhancement:
  • LoadRunner accepts non-USGS input files, for either/or qwdata or dv files. See help files for how to prepare data.
• Fix:
  • LoadRunner stopped copying out model selection data correctly into the individual site model.txt files, ever since added the allsite-summary file for model data. Fixed.
  • Shortened column header labels for allsite model summary, so that column data lines up with headers on the LoadRunner transcript display.

LoadRunner v 1.0
September 25, 2007

• Fix:
  • Made LoadRunner understand latest USGS file formats, plus the prior two formats since LoadRunner was begun only a year ago...

    USGS changed their water quality data (qwdata) file format to three new choices of format, all more verbose than the old, and changed the column names for the quantities of interest:

    1. "One sample per row with remark codes combined with values" (new default) : They added a slew of fields of no interest, then columns such as "p00410" with entries such as "48.6" and "E 53" and "< .1", combining the measurement for element code 00410 and its remark column into one field.
    2. "One sample per row with tab delimiter for remark codes" : same as the above, except adds columns such as "r00410" and tab-separates the remark ("E", "<") from the measurement ("48.6", "53").
    3. "One result per row" : Most similar to previous qwdata formats, this has only one element/measurement per row (such as 00410). However, they still added a lot of new uninteresting fields, and renamed the column that contains the measurement, so LoadRunner didn't understand this file format, either.
  • Note: Although LoadRunner's output is equivalent under the new file formats, some of the labeling won't match old runs. In the case where two elements of interest have measurements on the same day, LoadRunner averages their values and labels the calibration file entry with only one of the element codes, not both/all that went into the average. Then these munged lists of codes are used to report which elements were found at the site. Due to differences in processing, those comments may differ from old file formats, and between new formats. The resultant numbers, though, are the same.
  • LoadRunner's web home moved to http://environment.yale.edu/raymond/loadrunner/ .

LoadRunner v 0.9
May 24, 2007

• Enhancement:
  • Added "< detect" option. The choices are "drop" and "use 1/2". Drop was the previous behavior - all nondetects (coded in USGS qwdata files as ".1 <" with the .1 being the detection limit) were simply omitted from the LOADEST calibration data. The new option of "use 1/2" would code this value as ".05" with a caveat of "<". Drop and half detect values are both noted in the transcript. Note: sigma is NOT applied to quality values with a "<" caveat. And these caveat "<" values are not presented as nondetects to LOADEST - they are coded as normal values.
  • Moved LoadRunner to its new web home environment.yale.edu/loadrunner.

LoadRunner v 0.8b
May 10, 2007

• Enhancement:
  • Added "Loadrunner homepage" to list of help files, externally configured by the file loadrunner/loadrunnerconfig.properties. That way can put the help files outside of an installer, making them easier to maintain and update for users who've installed loadrunner.

LoadRunner v 0.8a
March 13, 2007

• Enhancement:
  • Added new option "flowGap" - maximum number of consecutive days to interpolate flow values (default 7 days max). Days skipped because of this option, simply don't appear in loadest's "est.inp" file, or loadrunner's daily flux output. Because months and years report averages of the values that are present, these don't change (though occasionally a month may be missing.)

LoadRunner v 0.8
March 6, 2007

• Enhancements:
  • Added allsite_elem_model.txt output file, reporting site/model/.../a0..aN statistics for each site in the run.

    This new allsite model report is the last thing dumped to the run transcript, and provides an easier way of telling how many sites completed during a LoadRunner job.

  • Changed all xxx.tsv reports to xxx.txt, since Access can't open Excel's .tsv files.
  • Added switches for preGapMin and postGapMin, the minimum number of calibration points for the first pre-gap group or last post-gap group. So if preGapMin is 3, and there are only 2 calibration points before the first gap, those 2 points get dropped. Likewise for the last group of points after the last gap. This is only applied once. So if the points go 2-gap-1-gap-20-gap-1-gap-2, only the first 2 and last 2 would get dropped by preGapMin = postGapMin = 3. The next inward groups of 1 point don't get dropped.

    The default is "Any" - which means beginning and ending groups of 1 calibration point are allowed.

    It is possible to run out of calibration points this way, so that the site won't run.

    HR03 (SAtlanticGulf) batch 5 has a number of sites where preGapMin=postGapMin=3 will prune off calibration points on alkalinity.

• Bugs fixed:
  • The dv (flow) file values, even within a dv file, switch back and forth between one and three fields for the flow value. In the 3-field version, the third field is mean flow (the one we want). Changed to correctly parse these column headers and read either column format, switching within the dv file.
  • The interpolation of the last entries in the estimated flow file sometimes ended up with invalid (-1) flows when available flow data ended before the end of the year. Those LOADEST jobs failed (LOADEST requires flows all be > 0). So more sites correctly run now. (The correct interpolation is to take the last valid flow and fill out the rest of the year with it.)

LoadRunner v 0.7e
February 22, 2007

• Enhancement:
  • Omit "lonely" starting calibration points - where the first calibration point is separated by more than maxGap from the second. These lonely points caused significantly raised load estimates on the all-contiguous-US alk runs.

LoadRunner v 0.7d
February 20, 2007

• Bugs fixed:
  • Very rarely, LoadRunner screwed up the matchDates process. This happened when cross-pruning landed the "common" list of years into a gap in one file or the other. Fixed part-way through running sulfates for contiguous US. (So had already done all doc and alk, and this is rare enough that we won't redo, probably.)

LoadRunner v 0.7c
February 5, 2007

• Bugs fixed:
  • LoadRunner stopped, no warnings, no complaints, halfway through HR01NewEngland on alk, so quietly that I never realized it had failed. Determined it was running out of memory along the way, so added "java -mx500M -cp..." option in the loadestRunner spawning. It should never need more than 500Mb of memory... Or at least, it isn't likely to get it anytime soon. (The default seemed to be 20 Mb per process... not a lot for 200 Mb input files.)
  • In 0.7b (which I never gave to anybody), it also dropped leading 0's on site number from the echo of the allsite_calib.tsv file (but not for the individual sites's calib.tsv files...)

LoadRunner v 0.7b
February 1, 2007

• Bugs fixed:
  • USGS inserted a new column "coll_ent_cd" to the qwdata file, so LoadRunner didn't read the columns correctly anymore. New version reads both old and new columns. Unless they change the column names, hopefully the way I fixed it will remain fixed even if they insert another column... to a point.

LoadRunner v 0.7a
January 10, 2007

• Bugs fixed:
  • Simply stopped (crashed) when had 0 quality or flow measurements for a site.
  • Left a lot of 0 length files and directories of 0 length files for sites that didn't run. Now if there were only a few calib data points, it outputs that to the site_elem_calib.tsv file under outputs, but deletes the loadest directory. If the site had no calib data points, cleans up all associated empty files.
  • Loadest file est.inp had list of all sites, instead of the current one.
  • The Stop button, if pressed before the job progressed to creating the loadest output directory, went away and failed to stop the job. Now the button stays and you can keep pressing it until it works....
  • Added sigma Off option, to skip the sigma check altogether. Running with sigma 3.0 is a good way to surf the data for anomalies, but having considered them, often the best next step is to just turn the sigma check off.
  • Help files didn't work in released version...

LoadRunner v 0.7
January 8, 2007

• Bugs fixed:
  • Omit calibration data points with concentration value of 0.0 or "censored" (less than detection limit) as these caused crash problems on LOADEST.
  • Accept spaces in any file/directory pathnames (inc. installation directory, run directory, inputs.)
  • Was replacing flow values < 1 with a value of 1.0. Changed to simply toss the value as bad and replace it with an interpolated value.
  • Average multiple concentration values for one day into one entry for the calibration file.
  • Failed to show last model output button and dump to results when ran only one site.
• Enhancements:
  • New site_elem_calib.tsv and allsite_elem_calib.tsv output files - the calib.inp files re-output to match other outputs. Note the caveat column is worthy of careful scrutiny. Also note that sites which didn't run have site_elem_calib.tsv files, but those entries aren't added to the allsite_elem_calib.tsv file. In other words, the allsite composite only includes calibration data that ran. (The most common reason a site didn't run, is that it didn't have enough calibration data.)
  • New menu bar with Exit and Help menus. Wrote all help files.
    • Most important is the "Element Codes" help file - can copy/paste lists of codes from there into ElemNum input field.
    • The "LoadRunner" help file is essentially the user manual.
    • This "Changes" help file replaces the relnotes.doc file from now on.
    • The "Acknowledgements" help file is still just a stub - should list references and funding, or whatever.

LoadRunner v 0.6
November 20, 2006

• Bug fix:
  • Gets more calibration points (and more runnable sites). Previous versions failed to read calibration points (from qual qwdata file) if there was a TabTab instead of TabValueTab in the time field.
• gapYear% added
  • This option allows you to specify the maximum number of years permitted in a “gap” (years with no calibration points), as a percentage of the years in a run. For example, if gapYear% is 10, and you have data from 1967 to 1986, the maximum gap would be two years. Then if years 1969 and 1970 were missing calibration points (gap 2), they would be included in the output. But if years 1980-1983 were missing (gap 4), those years are not included in the output.
• Speed:
  • Changed LOADEST header options to do less precise statistics, for a major speedup. (One hour to one minute on bicarb 49 site run.)

LoadRunner v 0.5
November 15, 2006

• Changed system for building calib.inp LOADEST input file, to use flow data from the qual file, if there is any. That means it:
  • Reads 00060 and 00061 records (flow and instantaneous flow)
  • Averages them if more than one flow record on the same day – there is no caveat for taking this average (having both is not noteworthy)
  • Uses those to provide flow information to go with concentration measurements for writing the calib.inp calibration file
  • Only if no flow value is found from the qual (qwdata) file, is flow taken from the flow (dv) input file. There is a caveat when this is done.
• Flow values for the est.inp LOADEST input file are unchanged – they come directly from the flow (dv) input file.
• Note that sometimes the USGS site spits out 3 tab-separated columns for one value (streamflow). LoadRunner can’t cope with that – it expects flow in the first column past the date. To get around this, text-edit the input (dv, flow) file by replacing that collection of tabs, to be date, tab, value.

LoadRunner v 0.4
November 6, 2006

• Added site column to all flux output files. Note that if splitSite is false (sites are merged), there is a comma-separated list of sites in this column.

LoadRunner v 0.3
November 2, 2006

• Builds allsite_elem_xxxflux.tsv summary files, including placing those on buttons. (There is no allsite_elem_model.txt report, though.) The allsite summaries are for daily, monthly, and annual, though site runs combine annual and monthly into one file. So, not very consistent anymore, but…
• Fixed rather mystifying termination (run ended happily, but with no output) when LoadRunner encountered blank lines at the ends of input files.

LoadRunner v 0.2
Halloween, 2006

• Added a STOP button on LoadRunner which goes and kills the LOADEST processes.
• Added “splitSite” option, true is the default. When split site is on, LoadRunner (actually loadestRunner) creates a bunch of individual LOADEST runs, one per site. The LoadRunner GUI shows you how it’s going with that. As each site finishes, the output buttons move to the latest outputs available.
• Made a LOT of changes to allow that splitSite thing – hopefully I’ve gotten everyt crash that prevents it from going on to the next run…
• Made bunch of requested changes to flux output files.
• Added actual constituent numbers used (rather than the ones I was told to look for) in the calib.inp files. So each entry is labeled with a constituent code, and the header gives a list of all the constituent codes that were used in this calib.inp.
• LOADEST simply stops when it runs into a flow value of 0, which is different from the invalid value of -1. I decided to replace those 0’s with 1’s and added a caveat code for that.
• There are also qwdata entries with a code of M = “presence verified but not quantified.” This is different than “

• The thing’s no speed demon. My 49 site bicarbonate test ran 1 hour 10 min.
• On the release directory under “loadrunner/examples” are a sitelist and notes on running the Ohio River project. The sitelist now contains 53 sites (the bicarb example I ran had only the first 49 of them.) The notes have the constituent numbers formatted as you need to enter them in the LoadRunner text box. Some of them are rather long lists and tedious to text-edit.
  • Ex: for “Ca”, you’d enter “00915,00916” (without the quotes)