potential divide by zero in support/scripts/size-stats

Message ID CAAkeqgUXF5eVOjQEUkM0OARw+cvMxGG-REv2rw-3BRu9y-yz0A@mail.gmail.com
State Changes Requested
Headers show
Series
  • potential divide by zero in support/scripts/size-stats
Related show

Commit Message

Andrey Yurovsky Nov. 12, 2017, 2:17 a.m.
I see a divide by zero exception in support/scripts/size-stats when I
use the graph-size target, the package of size zero causing this is
skeleton-init-systemd which I worked out by catching the exception. Is
this package erroneously of size zero or should that be OK?

---
 support/scripts/size-stats | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

Comments

Thomas Petazzoni Nov. 27, 2017, 9:08 p.m. | #1
Hello,

On Sat, 11 Nov 2017 18:17:36 -0800, Andrey Yurovsky wrote:
> I see a divide by zero exception in support/scripts/size-stats when I
> use the graph-size target, the package of size zero causing this is
> skeleton-init-systemd which I worked out by catching the exception. Is
> this package erroneously of size zero or should that be OK?

Thanks for reporting this bug and proposing a fix! I have some comments
on the implementation, but first about the format: the commit log
should contain a Signed-off-by line with your name and e-mail, and the
commit log should not be formatted as a personal message, but rather as
a description of what is being changed/fixed. Also the first line
should look like: "support/scripts/size-stats: ..."

> diff --git a/support/scripts/size-stats b/support/scripts/size-stats
> index af45000..3cd6838 100755
> --- a/support/scripts/size-stats
> +++ b/support/scripts/size-stats
> @@ -178,9 +178,12 @@ def gen_files_csv(filesdict, pkgsizes, outputf):
>                       "File size in system (%)"])
>          for f, (pkgname, filesize) in filesdict.items():
>              pkgsize = pkgsizes[pkgname]
> -            wr.writerow([f, pkgname, filesize, pkgsize,
> -                         "%.1f" % (float(filesize) / pkgsize * 100),
> -                         "%.1f" % (float(filesize) / total * 100)])
> +            try:
> +                wr.writerow([f, pkgname, filesize, pkgsize,
> +                        "%.1f" % (float(filesize) / pkgsize * 100),
> +                        "%.1f" % (float(filesize) / total * 100)])
> +            except ZeroDivisionError:
> +                print('WARNING: \"%s\" is size 0!' % pkgname)

I don't think we want to skip files from zero-sized packages: we want
to see them in the CSV. So perhaps something like:

		if pkgsize != 0:
			percent_pkg = float(filesize) / pkgsize * 100
		else:
			percent_pkg = 0

		percent_total = float(filesize) / total * 100

		wr.writerow([f, pkgname, filesize, pkgsize,
			     "%.1f" % percent_pkg,
			     "%.1f" % percent_total])

Could you test it, and resubmit as a new patch, with your Signed-off-by?

Thanks a lot!

Thomas Petazzoni

Patch

diff --git a/support/scripts/size-stats b/support/scripts/size-stats
index af45000..3cd6838 100755
--- a/support/scripts/size-stats
+++ b/support/scripts/size-stats
@@ -178,9 +178,12 @@  def gen_files_csv(filesdict, pkgsizes, outputf):
                      "File size in system (%)"])
         for f, (pkgname, filesize) in filesdict.items():
             pkgsize = pkgsizes[pkgname]
-            wr.writerow([f, pkgname, filesize, pkgsize,
-                         "%.1f" % (float(filesize) / pkgsize * 100),
-                         "%.1f" % (float(filesize) / total * 100)])
+            try:
+                wr.writerow([f, pkgname, filesize, pkgsize,
+                        "%.1f" % (float(filesize) / pkgsize * 100),
+                        "%.1f" % (float(filesize) / total * 100)])
+            except ZeroDivisionError:
+                print('WARNING: \"%s\" is size 0!' % pkgname)


 #