[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [equinox-dev] normalized jar


Which version of the jarprocessor are you using?

There was a bug (https://bugs.eclipse.org/bugs/show_bug.cgi?id=226850) fixed in 3.4.  This was leading to verification errors on the META-INF/eclipse.inf files.  Though if I remember correctly, it could have potentially led to different pack effort levels used on the pack step compared to the conditioning step which might affect .class files.

The org.eclipse.equinox.p2.jarprocessor contains a Verifier class which will unpack pack.gz files and verify their signatures.  It has a static main method that you can run.
If you have the conditioned jars from before packing, they should contain an eclipse.inf file containing "pack200.conditioned=true".





Janet Dmitrovich <jdmitrov@xxxxxxxxxx>
Sent by: equinox-dev-bounces@xxxxxxxxxxx

10/07/2008 05:55 PM

Please respond to
Equinox development mailing list <equinox-dev@xxxxxxxxxxx>

To
Equinox development mailing list <equinox-dev@xxxxxxxxxxx>
cc
Subject
Re: [equinox-dev] normalized jar





Hi thanks for the reply Andrew.

"If you are getting differences after unpack, you may not actually be using pack200 conditioned/normalized jars, or something went wrong in that -repack normalization step.
"

So I am finding differences in the jar sizes ( pre and post pack )


I'm fairly certain that I am using pack200 conditioned/normalized jars, since they were built with the "-repack" option.
>> Is there some way to validate this?

>>What type if things could go wrong in the -repack/ normalization step.

Janet Dmitrovich
WPLC Expeditor Software Development
jdmitrov@xxxxxxxxxx
512-838-4912 T/L:678-4912 FAX:512-838-3703
11501 Burnet Road, Austin TX 78758 (Internal ZIP: 9372)



Inactive hide details for Andrew Niefer ---10/07/2008 04:24:57 PM---The contents of the jar should be bit-wise the same, so theAndrew Niefer ---10/07/2008 04:24:57 PM---The contents of the jar should be bit-wise the same, so the only difference between pre & post pack (for a previously condition
Andrew Niefer <aniefer@xxxxxxxxxx>
Sent by: equinox-dev-bounces@xxxxxxxxxxx

10/07/2008 04:18 PM

Please respond to
Equinox development mailing list <equinox-dev@xxxxxxxxxxx>



To

Equinox development mailing list <equinox-dev@xxxxxxxxxxx>

cc

Subject

Re: [equinox-dev] normalized jar





The contents of the jar should be bit-wise the same, so the only difference between pre & post pack (for a previously conditioned jar), if any, would be in the format of the jar itself. Differences could be, for example, in size & crc information for a given zip entry appearing before or after the entry itself. I'm not sure that these differences would amount to a size difference for the jar.


In the case of nested jars which are checked against their containers, we do need the jars to be bitwise the same even in jar format. For this reason, the jarProcessor in eclipse does an additional "normalization" step which is different from the pack200 -repack conditioning.


If you are getting differences after unpack, you may not actually be using pack200 conditioned/normalized jars, or something went wrong in that -repack normalization step.


-Andrew

Janet Dmitrovich <jdmitrov@xxxxxxxxxx>
Sent by: equinox-dev-bounces@xxxxxxxxxxx

10/07/2008 03:58 PM

Please respond to
Equinox development mailing list <equinox-dev@xxxxxxxxxxx>


To
Equinox development mailing list <equinox-dev@xxxxxxxxxxx>
cc
Subject
[equinox-dev] normalized jar







Should the size of the jar ( Normalized and signed jar ) be the same pre-packand post-unpack ?

Janet Dmitrovich
WPLC Expeditor Software Development
jdmitrov@xxxxxxxxxx
512-838-4912 T/L:678-4912 FAX:512-838-3703
11501 Burnet Road, Austin TX 78758 (Internal ZIP: 9372)

_______________________________________________
equinox-dev mailing list
equinox-dev@xxxxxxxxxxx

https://dev.eclipse.org/mailman/listinfo/equinox-dev
_______________________________________________
equinox-dev mailing list
equinox-dev@xxxxxxxxxxx

https://dev.eclipse.org/mailman/listinfo/equinox-dev
_______________________________________________
equinox-dev mailing list
equinox-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/equinox-dev