[ 
https://issues.apache.org/jira/browse/ARIES-1212?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14035642#comment-14035642
 ] 

Cristiano Gavião commented on ARIES-1212:
-----------------------------------------

I agree that subsystems should fall back to using the file extension name when 
there is no mime type.
I don't think there is a API to get the mime-type of a entry, when that entry 
is a uncompressed file. but you can get a file inside a compressed file that 
tags it as mimetype compatible archive.

Thought, there is a API that lets you check the "magic code" of a file using 
its input stream. For example, a zip file will always have "0x504b0304" value. 
This way we can filter the contents of a subsystem archive and only process 
those entries that are Compressed files (with or without the mimetype entry 
file)...

> RawSubsystemResource should use the specified MIME type to filter contents 
> instead its name
> -------------------------------------------------------------------------------------------
>
>                 Key: ARIES-1212
>                 URL: https://issues.apache.org/jira/browse/ARIES-1212
>             Project: Aries
>          Issue Type: Improvement
>          Components: Subsystem
>    Affects Versions: subsystem-1.0.1
>            Reporter: Cristiano Gavião
>
> The Subsystem spec states that:
> bq. A Subsystem is deployed as a Subsystem Archive (.esa) file. Subsystem 
> Archives are used to store Subsystems and optionally their resources in a 
> standard ZIP-based file format.
> bq. This format is defined in Zip File Format. Subsystems normally use the 
> Subsystem Archive extension of .esa but are not required to. However there is 
> a special MIME type reserved for OSGi Subsystems that can be used to 
> distinguish Subsystems from normal ZIP files. This MIME type is: 
> application/vnd.osgi.subsystem
> bq. The type is defined in IANA application/vnd.osgi.subsystem.
> Current implementation don't care about MIME type and uses the content file's 
> name to do the filtering before process contents. that should be avoided and 
> potentially will prevent the bug stated in [ARIES-1148].



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Reply via email to