commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From sebb <seb...@gmail.com>
Subject Re: [COMPRESS] Changeset ideas
Date Mon, 30 Mar 2009 11:34:37 GMT
On 30/03/2009, Christian Grobmeier <grobmeier@gmail.com> wrote:
> Hi,
>
>
>  > The current ChangeSet API allows for:
>  > + deletion of entries by name
>  > + addition of entries by ArchiveEntry and InputStream.
>  >
>  > This is fine as far as it goes, but I think it would be useful to add:
>  > + addition of entries by File
>  > + replacement of an existing named entry by File or Entry+InputStream
>
>
> Sounds good, esspecially the replacement.
>  About additions by file - we use only streams in the api, is it good
>  to start with File now?

I think it would be useful to add File to the putxxx() methods, but I
suggest we track that in a separate e-mail thread.

>  I think this is useful too, having in mind that I wanted to propose
>  some util classes which allow creating zip files on File basis.
>
>
>  >
>  > It may also be useful to allow the location of new entries to be
>  > specified. For example, one might want to add META-INF data at the
>  > front of an archive. It would be useful to specify the locations as:
>  > + start
>  > + end
>
>
> I am not sure about this. I cannot imagine a use case for that.

Start is definitely useful for things such as META-INF and NOTICE,
LICENSE, README files. Otherwise, the default should be end of file.

If I remember correctly, "ar" files have ordering requirements, but
these are complicated, so are probably best dealt with by the
appropriate utility on the OS.

>
>  > I think it would be quite easy to implement:
>  > + open output file, add any starting entries
>  > + for each input entry, either copy, skip or replace with new entry
>  > + at end of input, add any final entries.
>  > + close archive files
>  >
>  > This could be achieved with:
>  > + HashSet containing names to be deleted
>  > + HashMap containing new entries for existing names
>  > + 2 Lists for new entries.
>  >
>  > Using Hashes would avoid scanning the list.
>  > Also, I don't think any of the above would need to be updated during
>  > perform(), which would allow them to be re-used on another archive.
>
>
> That would be cool. Actually I was thinking about  your suggestion with using
>  a copy of the Set to operate on. This would resolve this too.

Indeed.

>
>  Christian
>
>
>  ---------------------------------------------------------------------
>  To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
>  For additional commands, e-mail: dev-help@commons.apache.org
>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
For additional commands, e-mail: dev-help@commons.apache.org


Mime
View raw message