Recompilation instructions. Most patches require recompilation of XSPEC. Please follow the instructions below if necessary

1. If you have compiled xspec for other machine architectures since you originally installed it, you will need to reconfigure the package for your current machine. To do this:
   • cd {distribution root}/src/BUILD_DIR
   • ./hmake distclean
   • ./configure

2. Set environment variables for development (here are the csh & variant instructions):
   • setenv LHEASOFT {distribution root}/{architecture}, where architecture is the architecture specific directory, such as Linux_2.2_i686 or SunOS_5.6_sparc
   • source $LHEASOFT/lhea-init.csh

3. If you build with local models make sure to setenv LMODDIR to your local model directory.
4. Change directory to {distribution root}/src/spectral/xspec/src

5. type "hmake", followed by "hmake install". If you cleaned the distribution, then this will recompile the xspec binary from the beginning (but not the libraries upon which it depends).

The help command produces warnings about long lines. Bug report added on Dec 2, 2003.

The fix is to replace $LHEASOFT/spectral/xspec/help/xspec.hlp

The save model command with no filename can create a file with a random name instead of savexspec.xcm. Our thanks to Marek Gierlinksi for reporting this. Bug report added on Dec 2, 2003. UPDATE Jan 8, 2003. The save command with no filename is now unable to open the output file. The fix has been updated to correct this. Thanks to Emrah Kalemci for spotting this problem.

The fix is to replace spectral/xspec/src/COMMAND/wrtpro.f and rebuild XSPEC.

The chain command seg faults for models with frozen parameters. Our thanks to Roderick Johnstone for reporting this. Bug report added on Dec 2, 2003. UPDATED Dec 5, 2003: Roderick also supplied fixes for unitialized variables which produced problems under Linux. FINAL (?) UPDATE Dec 8, 2003: The proposal parameters were not being generated successfully, leading to too many rejections.

The fix is to replace spectral/xspec/src/MCMC/dochn.f, spectral/xspec/src/MCMC/runchn.f, and rebuild XSPEC.

There are several LaTeX typoes included in the copy of the manual in the release tar file. Our thanks to Terry Gaetz for reporting this. Bug report added on Dec 3, 2003.

The fix is to replace spectral/xspec/manual/u_appe.tex, spectral/xspec/manual/u_chap1.tex, spectral/xspec/manual/u_chap5.tex, and spectral/xspec/manual/u_chap6.tex,

The chain command seg faults with gcc3.3 under Linux. Our thanks to Roderick Johnstone for diagnosing this and supplying the fix. Bug report added on Dec 4, 2003.

The fix is to replace spectral/xspec/src/MCMC/cghain.f and rebuild XSPEC.

If there are multiple datasets in use and one other than the last one is replaced by a dataset without a response file then xspec will crash. This can happen if RESPFILE is none and the user intends to use the response command to read the response. A workaround is to ensure that RESPFILE points to the response file required. Our thanks to Mariano Mendez for reporting this. Bug report added on Dec 5, 2003.

The fix is to replace spectral/xspec/src/RESPONSE/inresp.f and rebuild XSPEC.

The plot mcmc command doesn't work correctly if there are frozen parameters. Our thanks to Roderick Johnstone for spotting this. Bug report added on Dec 9, 2003.

The fix is to replace spectral/xspec/src/MCMC/mcplot.f and rebuild XSPEC.

Under some circumstances XSPEC can crash if the APEC files are not available (eg if APECROOT has been set incorrectly). Our thanks to Marc Audard for reporting this. Bug report added on Dec 10, 2003.

The fix is to replace spectral/xspec/src/functions/ldaped.f, spectral/xspec/src/functions/smdem2.f and rebuild XSPEC.

The addline and modid commands can't find their Tcl scripts.

The fix is to replace spectral/xspec/manager/tclIndex then copy this file to $LHEASOFT/../spectral/xspec/manager.

XSPEC sometimes seg faults when fitting using a table model with errors. Our thanks to Shikui Tang for reporting this. Bug report added on Jan 6, 2004.

The fix is to replace is to get the latest tar file and rebuild XSPEC.

The error option on the eqw command fails when there is more than one dataset. Our thanks to Glenn Morris for reporting this. Bug report added on Jan 12, 2004.

The fix is to replace spectral/xspec/src/MODEL_COMP/eqwdth.f and rebuild XSPEC.

XSPEC ignores any quality information associated with the background spectrum. This has not been a problem in the past because the source and background spectra have had identical bad channels. However, this is not necessarily the case for XMM RGS spectra. Our thanks to Glenn Morris for drawing our attention to the XMM document identifying this problem. Bug report added on Jan 17, 2004.

The fix posted on Jan 17 has been withdrawn. At the moment the only workaround is to modify the QUALITY vector of the source file to take into account the QUALITY vector of the background file. Glenn Morris has kindly supplied a bash script pha_qual_merge.bash to do this.

If you give the error option to eqwidth after the model component number then the option is not applied till the next time that the eqwidth command is given. The workaround is obviously to give the error option before the model component number. Our thanks to Roderick Johnstone for spotting this problem. Bug report added on Jan 17, 2004.

The fix is to replace spectral/xspec/src/MODEL_COMP/eqwdth.f, spectral/xspec/src/MODEL_COMP/rdewpr.f, and rebuild XSPEC.

The error command will calculate the confidence region for a parameter that is linked to a previous parameter. This is inefficient if the user wants to get the confidence regions of all parameters using the command error 1-**. Our thanks to Marc Audard for reporting this lacuna. Bug report added on Jan 26, 2004.

The fix is to replace spectral/xspec/src/FIT/sperrf.f and rebuild XSPEC.

The vnei and vgnei models segfault under Debian with gcc 3.3.2. Our thanks to Karoly Makonyi for reporting this. Bug report added on Jan 27, 2004.

The fix is to replace spectral/xspec/src/functions/ionsneqs.f and rebuild XSPEC.

The abund command fails when reading values from a file. Only the first 18 values are read out of the 30 required. Our thanks to Gloria Sala Cladellas for reporting this. Bug report added on Feb 2, 2004.

The fix is to replace spectral/xspec/src/functions/func.inc, spectral/xspec/src/functions/fgunc.f, spectral/xspec/src/functions/fpunc.f, spectral/xspec/src/COSMO/stabnd.f, and rebuild XSPEC.

There is a typo in the xmmpsf model which produces a spurious error message when using an image to calculate the PSF mixing. Our thanks to Glenn Morris for reporting this. Bug report added on Feb 9, 2004.
UPDATE 2/14/04. Glenn also points out that XMMPSF-DEC is not read if it is given in dd:mm:ss.s format.

The fix is to replace spectral/xspec/src/functions/xmmmc.f and rebuild XSPEC.

The ascac model fails on all systems due to a corrupted file and on Linux and OSX due to an unSAVEd array. Bug report added on Feb 20, 2004.

The fix is to replace spectral/xspec/src/functions/make_psf.f, spectral/xspec/src/functions/ea.f, and rebuild XSPEC then copy spectral/xspec/manager/allpsf.nosm.fits, spectral/xspec/manager/allpsf.sm2.fits to $LHEASOFT/../spectral/xspec/manager.

Cosmology information is not stored by save. This fix also adds cosmology information to the xautosav.xcm file and to the output of show. Thanks to Jeremy Sanders for reporting this. Bug report added on Mar 27, 2004.

The fix is to replace spectral/xspec/src/COMMAND/autpro.f, spectral/xspec/src/COMMAND/wrtinf.f, spectral/xspec/src/COMMAND/wrtpro.f and rebuild XSPEC.

The model command doesn't deal correctly with incorrect models which are only one character long. This can lead to a segmentation violation in some circumstances. Thanks to Marc Audard for point this out. Bug report added on Mar 27, 2004.

The fix is to replace spectral/xspec/src/MODEL_COMP/mtchmd.f and rebuild XSPEC.

A check for TLMIN4 is hard-coded into the response file reading. This will fail if the columns in the file are not in the standard order. Thanks to Warren Foche for spotting this. Bug report added on Apr 6, 2004.

The fix is to replace spectral/xspec/src/FILE/rrspog.f and rebuild XSPEC.

On some systems xspec can seg fault when reading a spectrum file. Thanks to Allyn Tennant for diagnosing it. Bug report added on Apr 21, 2004.

The fix is to replace spectral/xspec/src/FILE/rddtg1.f and rebuild XSPEC.

When doing an error command on multiple parameters it is possible under rare circumstances for the value of the minimum statistic and parameters to get overwritten in between running error on different parameters. This leads to spurious messages about new minima being found. Thanks to Stefano Bianchi and Francesco Massaro for reporting this. Bug report added on Apr 22, 2004.

The fix is to replace spectral/xspec/src/FIT/clcunc.f and rebuild XSPEC.

It is possible for the error command to return a confidence region limit that lies outside the parameter hard limits. Thanks to Roderick Johnstone for reporting this. Bug report added on Apr 27, 2004.

The fix is to replace spectral/xspec/src/FIT/clcunc.f and rebuild XSPEC.

xspec will attempt to fit data even when there is no response present. This can lead to bizarre results and possible seg faults. Thanks to Allyn Tennant for reporting this. Bug report added on Apr 28, 2004.

The fix is to replace spectral/xspec/src/COMMAND/xscmnd.f and rebuild XSPEC.