[Scribo] Backup feature

Hi all,

I'm currently testing the squeak version of Scribo.
As soon as I can, I will put the code on the repository, with public
access rights, so everyone who wants to help will be welcome.

But first I need to make the unit tests pass!

There is a backup feature in Scibo (class Repository). But the VW code
uses UnixProcess and BOSS. As in Aida, I thing we could use
FileStream>>fileOutClass:andObject: instead of BOSS. UnixProcess is used
to make a tar file. But in Squeak we need OSProcess, a separate package.

I tried the tar file, and the size is almost the same.

So I think we could remove methods using UnixProcess, and simply use
binary files. Also renaming #bossFromFileSite: to #backupFromFileSite:
and #bossToFIle to #backupToFile would make sens (in both VW and Squeak
version of course).

What do you think ?

Cheers!

Nicolas
--
Nicolas Petton
http://nico.bioskop.fr
            ___
          ooooooo
         OOOOOOOOO
        |Smalltalk|
         OOOOOOOOO
          ooooooo
           \   /
            [|]
--------------------------------
Ma clé GPG est disponible ici :
http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xE788C34D

_______________________________________________
Aida mailing list
[hidden email]
http://lists.aidaweb.si/mailman/listinfo/aida

On Sat, 19 Jan 2008 08:31:36 +0100
Nicolas Petton <[hidden email]> wrote:

> There is a backup feature in Scibo (class Repository). But the VW code
> uses UnixProcess and BOSS. As in Aida, I thing we could use
> FileStream>>fileOutClass:andObject: instead of BOSS. UnixProcess is used
> to make a tar file. But in Squeak we need OSProcess, a separate package.

Assuming I used this in a production environment, I'd probably
set up something like logrotate to archive several versions of
backups. So all I'd require from the base system is a simple fileout
of whatever is necessary to restore operation quickly.

There is really no need to introduce OS dependencies because of a
minor feature like compressing a saved file. Every system you'd run
Scribo on has a scheduling tool and schedulable compression facilities.

Then again, I can get 500 GB SATA hard disks for 100 €, so the point
becomes even less relevant.

s.
_______________________________________________
Aida mailing list
[hidden email]
http://lists.aidaweb.si/mailman/listinfo/aida

Backup of Scribo website is not only backup of image or binary fileout
of website, but files like attachments on the filesytem too! So wee need
to have kind of .zip or .tar.gz support in Scribo, and portable, between
dialects and between OS-ses.

Also, I'm introducing an XML import/export of Documents. Idea is to have
all content on a filesystem, with XML "manifest" and content with
attachments in separate files, in a separate directory for each document.

This way a Scribo can be easily extended to a full Document Management
System with a portable content and specially, connectable to the archive
systems! This is very important for as in countries which  have an
e-docs archive law already in place.

BTW: document is a main entity in Scribo, representing any kind of
content, from wiki page to a classic document.

Janko


Stefan Schmiedl wrote:
--
Janko Mivšek
AIDA/Web
Smalltalk Web Application Server
http://www.aidaweb.si
_______________________________________________
Aida mailing list
[hidden email]
http://lists.aidaweb.si/mailman/listinfo/aida

On Sat, 19 Jan 2008 11:14:19 +0100
Janko Mivšek <[hidden email]> wrote:

> Also, I'm introducing an XML import/export of Documents. Idea is to have
> all content on a filesystem, with XML "manifest" and content with
> attachments in separate files, in a separate directory for each document.

Like OpenOffice does, good idea. Take it one step further and export
locally generated content as OpenOffice-document? That makes no sense
if the content is just a reference to a Word-document, but for
wiki-pages etc. it would sure be nice. Especially if you then go the
other way round and import oowriter or oocalc files into web-editable
scribo data :-)



Still, I'm not quite clear on how you propose to perform the backup.

If "everything" including current metadata is already in the filesystem,
I can suspend Scribo, run archiver, resume Scribo. I'd guess the
suspend/resume becomes necessary to avoid backing up inconsistent data.

If Scribo needs to export data from image-based "short-term" storage to
the filesystem, the fastest way would be suspend, export, archive,
resume.

Still the archiving is done completely outside of the web serving
application. Split responsibilities.

How about the following approach:
  wget http://scribo-host/prepare-for-backup
    returns when the filesystem data is ready for backup
  tar ...
  wget http://scribo-host/resume-after-backup

This requires no knowledge about calling OS-specific processes.
On Windows, you could use some VBScript or even resort to simple
batch-files if you don't bother closing two IE windows in the morning.

s.
_______________________________________________
Aida mailing list
[hidden email]
http://lists.aidaweb.si/mailman/listinfo/aida

Hi all,

How about using SIXX?
http://www.mars.dti.ne.jp/~umejava/smalltalk/sixx/index.html

It runs on VW and Squeak. It also supports many hooks for customizing
XML format.
(And I've also used it for saving my Wiki data).

2008/1/19, Janko Mivšek <[hidden email]>:
Cheers,
--
[:masashi | ^umezawa]
_______________________________________________
Aida mailing list
[hidden email]
http://lists.aidaweb.si/mailman/listinfo/aida