Re: Checksum Format for .md5 and .sha1 Files

2009-01-02 Thread Brian Fox
I'm -1 to making a new format, but ambivalent about changing it to  
match openssl standards


--Brian (mobile)


On Jan 2, 2009, at 5:31 AM, Benjamin Bentmann  
 wrote:



Hi,

With regard to MINSTALL-47, I would like to discuss if we can/should  
change the format used for checksum files generated by Maven.


Currently, Maven outputs the plain checksum as-is to the checksum  
file (cf. [1]), e.g.


1dbbdec30feca97198d5d7e96d42dc57

In the mentioned issue, it was requested to append the filename,  
separated by a space and asterisk, to the checksum:


1dbbdec30feca97198d5d7e96d42dc57 *maven-metadata.xml

My naive assumption is that any decent checksum verification tool  
out there is able to strip extraneous characters from the checksum  
file if not processed. In contrast, guessing missing bits is usually  
harder for tools. So I believe changing the checksum format as  
requested would increase tool support.


Maven itself can already process this checksum format (due to  
MNG-797) and an IT is online (for MNG-1349 which merely requires  
newer Maven versions for SHA-1 checksums in OpenSSL format).


Technically, this change would affect the Maven Install Plugin and  
the DefaultWagonManager from maven-artifact-manager (aka maven- 
compat).


WDYT?


Benjamin


[0] http://jira.codehaus.org/browse/MINSTALL-47
[1] 
http://repo1.maven.org/maven2/org/apache/maven/plugins/maven-metadata.xml.md5

-
To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
For additional commands, e-mail: dev-h...@maven.apache.org



-
To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
For additional commands, e-mail: dev-h...@maven.apache.org



Re: Checksum Format for .md5 and .sha1 Files

2009-01-02 Thread Benjamin Bentmann

Brian Fox wrote:


I'm -1 to making a new format


Just to make sure we all have the same understanding: The proposed 
format is not "new" as in "yet another checksum format". It's an already 
existing format used by the md5sum tool (compare the format attribute of 
Ant's checksum task [0]).



Benjamin


[0] http://ant.apache.org/manual/CoreTasks/checksum.html


-
To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
For additional commands, e-mail: dev-h...@maven.apache.org



Re: Checksum Format for .md5 and .sha1 Files

2009-01-02 Thread Tamás Cservenák
Not to mention, that there is already a lot of files generated by md5sum and
sha1sum apps on central:
http://repo2.maven.org/maven2/log4j/log4j/maven-metadata.xml.md5
http://repo2.maven.org/maven2/log4j/log4j/maven-metadata.xml.sha1

But in the above cases, the path is obviously misleading.

~t~

On Fri, Jan 2, 2009 at 5:36 PM, Benjamin Bentmann  wrote:

> Brian Fox wrote:
>
>  I'm -1 to making a new format
>>
>
> Just to make sure we all have the same understanding: The proposed format
> is not "new" as in "yet another checksum format". It's an already existing
> format used by the md5sum tool (compare the format attribute of Ant's
> checksum task [0]).
>
>
> Benjamin
>
>
> [0] http://ant.apache.org/manual/CoreTasks/checksum.html
>
>
>
> -
> To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
> For additional commands, e-mail: dev-h...@maven.apache.org
>
>


-- 
Thanks,
~t~


Re: Checksum Format for .md5 and .sha1 Files

2009-01-02 Thread Oleg Gusakov
I remember that wagon, at least 1.0-beta-4, strips everything after " *" 
in the signature file including, so it should already be implemented on 
the reading side. Wagon team - please correct me if I am wrong - I did 
not check out the wagon source.


So we'll have to add it to the writing side of wagon code and I'll add 
this to Mercury.


Question is - should verification fail if file name does not match? In 
the light of the latest event - CCC generating MD5 collisions - it's 
kind of pointless.


Overall - MD5 and SHA1 can only check for non-intentional data 
modifications, they don't address non-repudiation problems. And when was 
it last time you data was crippled during transfer? It does not happen 
any more..


That is why we should switch to PGP signatures as soon as possible - 
Mercury and Mercury Ant provide full support for that :)


Thanks,
Oleg

Tamás Cservenák wrote:

Not to mention, that there is already a lot of files generated by md5sum and
sha1sum apps on central:
http://repo2.maven.org/maven2/log4j/log4j/maven-metadata.xml.md5
http://repo2.maven.org/maven2/log4j/log4j/maven-metadata.xml.sha1

But in the above cases, the path is obviously misleading.

~t~

On Fri, Jan 2, 2009 at 5:36 PM, Benjamin Bentmann   

wrote:



  

Brian Fox wrote:

 I'm -1 to making a new format

Just to make sure we all have the same understanding: The proposed format

is not "new" as in "yet another checksum format". It's an already existing
format used by the md5sum tool (compare the format attribute of Ant's
checksum task [0]).


Benjamin


[0] http://ant.apache.org/manual/CoreTasks/checksum.html



-
To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
For additional commands, e-mail: dev-h...@maven.apache.org






  


Re: Checksum Format for .md5 and .sha1 Files

2009-01-02 Thread Benjamin Bentmann

Oleg Gusakov wrote:

I remember that wagon, at least 1.0-beta-4, strips everything after " *" 
in the signature file including, so it should already be implemented on 
the reading side. Wagon team - please correct me if I am wrong - I did 
not check out the wagon source.


The checksum verification actually happens in Maven itself, more 
precisely in DefaultWagonManager.verifyChecksum(), independently of the 
Wagon version. Most of this parsing code dates back to Maven 2.0-beta-1 
(MNG-797).



Benjamin

-
To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
For additional commands, e-mail: dev-h...@maven.apache.org