Re: [Lzip-bug] Lzip 1.20-pre2 released

[Timothy Beryl Grahek]

Hi Antonio,

On 09/20/2017 02:39 AM, Antonio Diaz Diaz wrote:
> Lzip 1.20-pre2 is ready for testing here
> http://download.savannah.gnu.org/releases/lzip/lzip-1.20-pre2.tar.lz
> http://download.savannah.gnu.org/releases/lzip/lzip-1.20-pre2.tar.gz
>
> The sha256sums are:
> e97b0677bb5152e9450bb0480d517c4ce658c58a042867df7729e745c20b0c70 
> lzip-1.20-pre2.tar.lz
> 6fb5de275d79268fbb0ec27c7a5c2be7bd783e5efe2da6535c22ec45632f2f68 
> lzip-1.20-pre2.tar.gz 

I used the sha256sum you generated on lzip-1.20-pre2.tar.gz before I 
installed it. I have tested the following and there are no bugs after 
testing each more than once:

   Inverse compression ratio has replaced bits/byte
   Progress is shown in decompression and testing with ‘vv’
   Dictionary size is shown in decompression or testing using ‘vvvv’,
   while it is absent using 'vvv'

Here is proof:

   $ lzip -vv0 Test.tar
   Test.tar:  2.470:1, 40.48% ratio, 59.52% saved, 158576640 in,
   64198343 out.
   $ lzip -vvd Test.tar.lz
   Test.tar.lz:  2.470:1, 40.48% ratio, 59.52% saved.  done
   $ lzip -vvvvt Test.tar.lz
   Test.tar.lz: dictionary   64 KiB,  2.470:1, 40.48% ratio, 59.52%
   saved.  CRC 39CABC26, decompressed 158576640, compressed 64198343.  ok
   $ lzip -vvvt Test.tar.lz
   Test.tar.lz:  2.470:1, 40.48% ratio, 59.52% saved.  decompressed
   158576640, compressed 64198343.  ok
   $ lzip -vvvvd Test.tar.lz
   Test.tar.lz: dictionary   64 KiB,  2.470:1, 40.48% ratio, 59.52%
   saved.  CRC 39CABC26, decompressed 158576640, compressed 64198343.  done

But then I think I have found a minor bug that is unrelated to the 
things I have tested. Here is what happened:

   $ lzip -vvvv0 Test.tar.lz
   lzip: Input file 'Test.tar.lz' already has '.lz' suffix.

This is a very informative message; however, it doesn't let me know 
whether this file or a file without the '.lz' suffix exists or not. I 
don't know if this ultimately matters, but I just thought I would 
mention it. Accordingly, I think it would be helpful if Lzip could 
indicate that any file with or without the '.lz' suffix does or does not 
exist, such as like this:

   $ lzip foo.tar.lz
   lzip: Input file 'foo.tar.lz' already has '.lz' suffix. 'foo.tar.lz'
   exists.

   $ lzip foo.tar.lz
   lzip: Input file 'foo.tar.lz' already has '.lz' suffix. 'foo.tar'
   exists.

   $ lzip foo.tar.lz
   lzip: Input file 'foo.tar.lz' already has '.lz' suffix. 'foo.tar'
   does not exist.

It seems reasonable to mention the existence of 'foo.tar.lz' only if 
'foo.tar' doesn't exist; if 'foo.tar' does exist, it's not meaningful to 
mention that 'foo.tar.lz' also exists. Similarly, if neither 'foo.tar' 
exists nor 'foo.tar.lz' exists, then it's reasonable to me to mention 
that 'foo.tar' does not exist, while it is meaningless to mention 
anything about 'foo.tar.lz'.

Anyhow, do you think this is a worthwhile feature to implement? If not, 
I completely understand, but I just thought I would bring it to your 
attention.

I will use Lzip 1.20 Pre2 on all my machines that I can install it on 
and I will look for any other bugs. The results posted above are based 
on Arch Linux x86-64. I had some problems installing Lzip from source 
the other day using Cygwin; if I still keep having problems with that, I 
will open a thread about it; it's probably unrelated to the new release 
and is likely just a Cygwin problem.

Best regards,

Timothy Beryl Grahek