sage -t tests/cmdline.py creates junk file logs/pkgs/sqlite.log

[seblabbe]

As reported on sage-release:

rm -f logs/pkgs/sqlite.log
sage -t src/sage/tests/cmdline.py
ls logs/pkgs/sq*

creates file

logs/pkgs/sqlite.log

... which is not a log file at all.

CC: @jdemeyer @slel @tornaria @mkoeppe

Component: scripts

Reviewer: Michael Orlitzky

Issue created by migration from https://trac.sagemath.org/ticket/22062

[jhpalmieri]

comment:1

I believe that the log file is created by the line

sage: out, err, ret = test_executable(["sage", "-p", "--info", "--info", "sqlite"])

Indeed, running sage --info sqlite just prints the SPKG.txt file to the screen, while sage -p --info sqlite also dumps that to a log file. This is because of the way the script src/bin/sage is written. We could change that:

diff --git a/src/bin/sage b/src/bin/sage
index cb3b611..3fb83ac 100755
--- a/src/bin/sage
+++ b/src/bin/sage
@@ -844,14 +844,19 @@ install() {
                 continue;;
         esac
 
-        PKG_NAME=`echo "$PKG" | sed -e "s/\.spkg$//"`
-        PKG_NAME=`basename "$PKG_NAME"`
-
-        sage-logger \
-            "sage-spkg $INSTALL_OPTIONS '$PKG'" "$SAGE_LOGS/$PKG_NAME.log"
-        # Do not try to install further packages if one failed
-        if [ $? -ne 0 ]; then
-            exit 1
+        if [[ $INSTALL_OPTIONS == *"-info"* ]]; then
+            # Just print the info, don't write a log file.
+            sage-spkg --info "$PKG" || exit $?
+        else
+            PKG_NAME=`echo "$PKG" | sed -e "s/\.spkg$//"`
+            PKG_NAME=`basename "$PKG_NAME"`
+
+            sage-logger \
+                "sage-spkg $INSTALL_OPTIONS '$PKG'" "$SAGE_LOGS/$PKG_NAME.log"
+            # Do not try to install further packages if one failed
+            if [ $? -ne 0 ]; then
+                exit 1
+            fi
         fi
     done
     # Display a message if we actually installed something (using this

[jhpalmieri]

comment:4

This might also be addressed by #30649, although I haven't tested that.

[jhpalmieri]

comment:6

This has indeed now been fixed, so I propose closing this.

[orlitzky]

Reviewer: Michael Orlitzky