record time, version in sage-starts

Question

record time, version in sage-starts

Closed this issue 14 years ago · 25 comments

The message printed by sage-starts should record the time and the version of Sage, especially since the output is now logged.

Apply attachment: trac_11991-sage-starts.v3.patch to the scripts repo.

Depends on #11926

Component: scripts

Author: John Palmieri

Reviewer: Leif Leonhardy

Merged: sage-4.8.alpha1

Issue created by migration from https://trac.sagemath.org/ticket/11991

Answer 1 · 2011-11-04T21:06:39.000Z

comment:1

Sample output (after running sage -sh):

$ ./sage-starts

Testing that Sage starts...
[2011-11-04 14:02:56] Sage Version 4.7.2.rc1, Release Date: 2011-10-27.  
Yes, Sage starts.

The last two lines are logged to start.log.

Answer 2 · 2011-11-04T23:36:42.000Z

comment:2

Ahem, what if Sage doesn't start? Then we get no date...

So you could either do something like

echo "[`date +'%Y-%m-%d %H:%M:%S'`]" \
     "`sage --version | sed -n -e '/Version/s/^[ |]\+//;s/[ |]\+$//p'`." \
    | tee -a "$SAGE_ROOT"/start.log

spkg/pipestatus "sage -c \"print 'Yes, Sage starts.'\" 2>&1" "tee -a '$SAGE_ROOT'/start.log"

or, directly,

echo "[`date +'%Y-%m-%d %H:%M:%S'`]" \
     "`sed -n -e '/Version/s/^[ |]\+//;/Version/s/[ |]\+$//p' "$SAGE_LOCAL"/bin/sage-banner`." \
    | tee -a "$SAGE_ROOT"/start.log

spkg/pipestatus "sage -c \"print 'Yes, Sage starts.'\" 2>&1" "tee -a '$SAGE_ROOT'/start.log"

(sage --version is currently broken w.r.t. the removal, so you could actually move the sed command to sage-sage.)

or use sage --python -c 'your favourite Python code' for printing the date and Sage's version (without importing sage.*).

Answer 3 · 2011-11-04T23:36:42.000Z

Reviewer: Leif Leonhardy

Answer 4 · 2011-11-05T00:00:32.000Z

comment:3

Instead of sage-banner (which should have a .txt extension btw.), you could also use one of our VERSION.txt files.

Answer 5 · 2011-11-05T00:21:30.000Z

comment:4

Replying to @nexttime:

Ahem, what if Sage doesn't start? Then we get no date...

Oops, of course you're right. Here's a new patch, using python instead of sage for the date and version. (Note that the file sage/version.py is extremely simple, so if Python functions, we should be able to import it.)

Answer 6 · 2011-11-05T00:21:48.000Z

scripts repo

Answer 7 · 2011-11-05T00:53:47.000Z

comment:5

Attachment: trac_11991-sage-starts.patch.gz

Replying to @jhpalmieri:

(Note that the file sage/version.py is extremely simple, so if Python functions, we should be able to import it.)

It is, but note that also sage/__init__.py is executed when you import sage.version.

This is not a problem right now, but may become one if Sage's import mechanics change (which is work in progress). You could perhaps modify PYTHONPATH and import version, but I haven't tried that yet^*. Maybe run Sage's Python directly then (instead of ./sage --python ...).

^* PYTHONPATH="devel/sage/sage" local/bin/python -c 'import version; ...' should work, both from within and outside a Sage environment.

Answer 8 · 2011-11-05T01:29:25.000Z

comment:6

Here's another version, using VERSION.txt. This will therefore record any update information stored there.

Answer 9 · 2011-11-05T04:05:42.000Z

comment:7

How about

version=spkg/standard/VERSION.txt
echo "[`date +'%Y-%m-%d %H:%M:%S'`] `cat $version 2>/dev/null`" | tee -a start.log
...

or

version=spkg/standard/VERSION.txt
(echo -n "[`date +'%Y-%m-%d %H:%M:%S'`] "; cat $version 2>/dev/null) | tee -a start.log
...

?

Answer 10 · 2011-11-05T05:33:39.000Z

comment:8

Sure, that's fine.

Answer 11 · 2011-11-05T07:03:11.000Z

comment:9

Replying to @jhpalmieri:

Sure, that's fine.

Well, if you prepend $SAGE_ROOT/ (which isn't necessary, since we've cd'ed to it), then you have to quote $version (with double quotes). ;-)

Answer 12 · 2011-11-05T15:20:45.000Z

Attachment: trac_11991-sage-starts.v2.patch.gz

scripts repo

Answer 13 · 2011-11-05T15:21:36.000Z

Description changed:

--- 
+++ 
@@ -1,2 +1,6 @@
 The message printed by sage-starts should record the time and the version of Sage, especially since the output is now logged.
 
+---
+
+Apply [attachment: trac_11991-sage-starts.v2.patch](https://github.com/sagemath/sage-prod/files/10653971/trac_11991-sage-starts.v2.patch.gz) to the scripts repo.
+

Answer 14 · 2011-11-05T15:21:36.000Z

comment:10

Another attempt (without $SAGE_ROOT).

Answer 15 · 2011-11-05T22:54:17.000Z

comment:11

Actually only depends on one part (the one to sage-starts) of one patch (attachment: ticket:11926:11926_sage_starts.patch) at #11926.

sage-starts could by the way be more robust, i.e., it shouldn't print the following if the current working directory just happens to not be $SAGE_ROOT, and SAGE_ROOT isn't set.

$ bin/sage-starts 

Testing that Sage starts...
[2011-11-05 23:15:43] 
bin/sage-starts: line 14: spkg/pipestatus: No such file or directory
Sage failed to start up.
Please email sage-devel (http://groups.google.com/group/sage-devel)
explaining the problem and send the log file
  /home/leif/Sage/sage-4.7.2-gcc-4.5.1/local/start.log
Describe your computer, operating system, etc.

(Also note the logfile's name / location. The log just contains the time stamp, and left-over files are likely to pollute repositories; only the root repository ignores *.log files. I'm not saying that this is something a user will typically run into [unless e.g. he/she includes $SAGE_LOCAL/bin into his/her PATH, which apparently some do, although they clearly shouldn't^*], but one shouldn't print a message that encourages people to submit useless error reports.)

Otherwise positive review, since that's not really related to the purpose of this ticket. Feel free to fix it here (or elsewhere) though.

^* We may add this to the Installation Guide (that one shouldn't put $SAGE_ROOT/local/bin/ into PATH), but IMHO most scripts there should do the "usual" sanity check, i.e. for example test whether SAGE_LOCAL is defined, since most (or at least many) depend on the Sage environment being already set up.

Answer 16 · 2011-11-05T22:54:17.000Z

Dependencies: #11926

Answer 17 · 2011-11-06T03:29:51.000Z

comment:12

We could add something like this:

diff --git a/sage-starts b/sage-starts
--- a/sage-starts
+++ b/sage-starts
@@ -7,6 +7,13 @@ if [ -n "$SAGE_ROOT" ]; then
     cd "$SAGE_ROOT"
 fi
 
+if [[ ! -x spkg/pipestatus ]]; then
+    echo >&2 "Error: The file spkg/pipestatus was not found or is not"
+    echo >&2 "executable, perhaps because the 'sage-starts' script was"
+    echo >&2 "not run from the \$SAGE_ROOT directory.  Exiting."
+    exit 1
+fi
+
 echo
 echo "Testing that Sage starts..."
 version=spkg/standard/VERSION.txt

Answer 18 · 2011-11-06T03:46:18.000Z

comment:13

Replying to @jhpalmieri:

We could add something like this:

diff --git a/sage-starts b/sage-starts
--- a/sage-starts
+++ b/sage-starts
@@ -7,6 +7,13 @@ if [ -n "$SAGE_ROOT" ]; then
     cd "$SAGE_ROOT"
 fi
 
+if [[ ! -x spkg/pipestatus ]]; then
+    echo >&2 "Error: The file spkg/pipestatus was not found or is not"
+    echo >&2 "executable, perhaps because the 'sage-starts' script was"
+    echo >&2 "not run from the \$SAGE_ROOT directory.  Exiting."
+    exit 1
+fi
+
 echo
 echo "Testing that Sage starts..."
 version=spkg/standard/VERSION.txt

I had exactly the same in mind, although I would just print: "This script has to be run from the SAGE_ROOT directory or with SAGE_ROOT properly set."

I won't mind if you add your suggestion to your patch. You could prepend `pwd`/ to spkg/pipestatus (in the message) and perhaps then put that filename on a line by its own.

Answer 19 · 2011-11-06T03:57:48.000Z

comment:14

Although we could be a bit more clever and move the test up, i.e.

if [[ -x "$SAGE_ROOT"/spkg/pipestatus ]]; then
    cd "$SAGE_ROOT"
elif [[ ! -x spkg/pipestatus ]]; then
    # Complain that either SAGE_ROOT isn't set properly (if it's non-empty)
    # or that we're [presumably] in the wrong directory (otherwise).
    exit 1
fi

Answer 20 · 2011-11-06T04:09:41.000Z

comment:15

s/Complain that either/Complain either that/, i.e.

    ...
    if [[ -z $SAGE_ROOT ]]; then
        echo >&2 "Error: This script ($0) has to be run from the SAGE_ROOT directory."
    else
        echo >&2 "Error: SAGE_ROOT(=$SAGE_ROOT) isn't properly set."
    fi
    ...

Answer 21 · 2011-11-06T05:10:29.000Z

comment:16

Here's version 3. If you think it's okay, please change the ticket description to refer to this patch instead of v2.

Answer 22 · 2011-11-06T05:10:41.000Z

Attachment: trac_11991-sage-starts.v3.patch.gz

scripts repo

Answer 23 · 2011-11-06T05:34:03.000Z

Description changed:

--- 
+++ 
@@ -2,5 +2,5 @@
 
 ---
 
-Apply [attachment: trac_11991-sage-starts.v2.patch](https://github.com/sagemath/sage-prod/files/10653971/trac_11991-sage-starts.v2.patch.gz) to the scripts repo.
+Apply [attachment: trac_11991-sage-starts.v3.patch](https://github.com/sagemath/sage-prod/files/10653972/trac_11991-sage-starts.v3.patch.gz) to the scripts repo.

Answer 24 · 2011-11-06T05:34:03.000Z

comment:17

Replying to @jhpalmieri:

Here's version 3. If you think it's okay, please change the ticket description to refer to this patch instead of v2.

Done and done.

As a side-effect, even SAGE_ROOT=/foo/bar local/bin/sage-starts now works. :)

Answer 25 · 2011-11-07T10:13:13.000Z

Merged: sage-4.8.alpha1