History log and timemachine, a proposal

[Cornelius_Weiss]

Hi devs,

i just worked out a proposal for history logging and a timemachine/
waybackmachine:
http://www.tine20.org/wiki/index.php/Developers/Egwbase/Timemachine

it would be great if some of you take a look at it and comment it, as
I want to start implementation of this stuff (for tine) between the
years.

cu
conny

---

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

---

eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers

[Jose_Luis_Gordo_Rome]

Hi Cornelius,

That’s great, automatic changelog!

This will be performance viable for large concurrent sites?

Another question

The tine team are designing with global search in mind? I think is an very
important feature that current egw doesn’t have, my user wants global search
because don’t remember if they put a task, a calendar reminder, tracker item
…

Thanks in advance

Hi devs,

i just worked out a proposal for history logging and a timemachine/
waybackmachine:
http://www.tine20.org/wiki/index.php/Developers/Egwbase/Timemachine

it would be great if some of you take a look at it and comment it, as
I want to start implementation of this stuff (for tine) between the
years.

cu
conny

---

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

---

eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers

–
Saludos
José Luis Gordo Romero

[Nathan_Gray-2]

Hi devs,

i just worked out a proposal for history logging and a timemachine/
waybackmachine:
http://www.tine20.org/wiki/index.php/Developers/Egwbase/Timemachine

it would be great if some of you take a look at it and comment it, as
I want to start implementation of this stuff (for tine) between the
years.

Egwbase_Timemachine_Interface::getRecord() takes a string,
Egwbase_Timemachine_Model_ModificationLog::record_identifier is an
int. I may be mis-reading your diagram, but this seems inconsistent.
Will this work for a multipart key? I get those a lot.

Nathan Gray
nathan at goarctic dot com

---

Extra punctuation??? Ug!!!
Just say no! Punctuation marks do not gain significance if you
increase their number, they make you look silly. One is sufficient.

---

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

---

eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers

[Ralf_Becker_6]

Hi José,

a global search is easy to implement, for the apps you talked, or for
all apps supporting the link interface. They all have a standard search
capability, we only lack a UI to use it. I wanted start that a view
times, but I got always distracted.

The search is over each single app, so the UI of the result page would
look like paragraphs for each apps eg. with bullet points for each found
entry displaying it’s title and link to view the entry.

My idea on a long run would be to start using views for the link-titles,
so you could make a join with the links table and these title views to
query all links of an entry. At the moment this is done with php code
via the business objects of each app. That’s for example the reason why
InfoLog (which displays all links) is a lot slower then eg. the
addressbook (which does not display them).

Ralf

José Luis Gordo Romero schrieb:

Hi Cornelius,

That’s great, automatic changelog!

This will be performance viable for large concurrent sites?

Another question

The tine team are designing with global search in mind? I think is an
very important feature that current egw doesn’t have, my user wants
global search because don’t remember if they put a task, a calendar
reminder, tracker item …

Thanks in advance

2007/12/21, Cornelius Weiss <c.weiss@metaways.de
mailto:c.weiss@metaways.de>:

Hi devs,

i just worked out a proposal for history logging and a timemachine/
waybackmachine:
http://www.tine20.org/wiki/index.php/Developers/Egwbase/Timemachine
<http://www.tine20.org/wiki/index.php/Developers/Egwbase/Timemachine>

it would be great if some of you take a look at it and comment it, as
I want to start implementation of this stuff (for tine) between the
years.

cu
conny


-------------------------------------------------------------------------

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
<http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/>
_______________________________________________
eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
<mailto:eGroupWare-developers@lists.sourceforge.net>
https://lists.sourceforge.net/lists/listinfo/egroupware-developers

–
Saludos
José Luis Gordo Romero

---

---

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

---

---

eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers

–
Ralf Becker
eGroupWare Training & Support ==> http://www.egroupware-support.de
Outdoor Unlimited Training GmbH [www.outdoor-training.de]
Handelsregister HRB Kaiserslautern 3587
Geschäftsführer Birgit und Ralf Becker
Leibnizstr. 17, 67663 Kaiserslautern, Germany
Telefon +49 (0)631 31657-0

---

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

---

eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers

[Cornelius_Weiss]

Hi Nathan,

you’re right, i made a mistake.

In Egwbase_Record_Abstract i was is the identifier property. In the
beginning i thought that i let this identifier of each record a mixed
datatype, either string or int, to have the possibility for e.g.
uid’s. However after studing the corrensponding rfc’s i think the
best is to have the local identifier as int, and to have a global uid
seperate from this.

I need to chage datatypes consitent through the classes.
thanks for the hint

as for the multipart key. I’m not shure. If you use them in db, you
also have to use them in the frontend. This could slow down things.
Therefore I think atm. that the best is, to allways maintain a
sequence or autoincrement int.

cu
conny

Hi devs,

i just worked out a proposal for history logging and a timemachine/
waybackmachine:
http://www.tine20.org/wiki/index.php/Developers/Egwbase/Timemachine

it would be great if some of you take a look at it and comment it, as
I want to start implementation of this stuff (for tine) between the
years.

Egwbase_Timemachine_Interface::getRecord() takes a string,
Egwbase_Timemachine_Model_ModificationLog::record_identifier is an
int. I may be mis-reading your diagram, but this seems inconsistent.
Will this work for a multipart key? I get those a lot.

Nathan Gray
nathan at goarctic dot com

---

Extra punctuation??? Ug!!!
Just say no! Punctuation marks do not gain significance if you
increase their number, they make you look silly. One is sufficient.

---

---

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

---

eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers

---

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

---

eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers

[Cornelius_Weiss]

Hi Jose,

Hi Cornelius,

That’s great, automatic changelog!

This will be performance viable for large concurrent sites?

Thats a good point. I could think of different solutions for big sites:

• move enties oder than xxx to a shadow table
• partition tables depending on age of entries (mysql only?)
• the two points above but not with time but with amount of history
  for one record

Another question

The tine team are designing with global search in mind? I think is
an very important feature that current egw doesn’t have, my user
wants global search because don’t remember if they put a task, a
calendar reminder, tracker item …
yes, not before milestone 3, but it’s an topic. Please take a look at:
http://framework.zend.com/manual/en/zend.search.lucene.html

Thanks in advance
shure

cu
conny

[Cornelius_Weiss]

argh, sorry, i answerd before i looked into the uml…

the Egwbase_Timemachine_Interface::getCUD functions returning uid’s /
array of uid’s. This is, becuase the usage of those functions is
mostly needed by syncinc stuff.

The Egwbase_Timemachine_Modification* is just internal stuff and
takes identifiers (int)

Maybe it would be best to have 2 seperate functions in
Egwbase_Timemachine_Interface one for indentifiers and one for global
uids

cu
conny

Hi devs,

i just worked out a proposal for history logging and a timemachine/
waybackmachine:
http://www.tine20.org/wiki/index.php/Developers/Egwbase/Timemachine

it would be great if some of you take a look at it and comment it, as
I want to start implementation of this stuff (for tine) between the
years.

Egwbase_Timemachine_Interface::getRecord() takes a string,
Egwbase_Timemachine_Model_ModificationLog::record_identifier is an
int. I may be mis-reading your diagram, but this seems inconsistent.
Will this work for a multipart key? I get those a lot.

Nathan Gray
nathan at goarctic dot com

---

Extra punctuation??? Ug!!!
Just say no! Punctuation marks do not gain significance if you
increase their number, they make you look silly. One is sufficient.

---

---

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

---

eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers

---

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

---

eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers

[Cornelius_Weiss]

Hi Nathan et all,

I just reworked the interfaces, and now they are hopefully consitent.

NOTE: ids (int) are used internally, uids (strings) are used when
dealing with external data (e.g. multiple calendars (external icals))
or sync or…

Here is the UML and description:
http://www.tine20.org/wiki/index.php/Developers/Egwbase/Timemachine

If there are not other ideas, i’ll start coding tomorrow.
I had to postpone it a bit as work on child 2.0 (http://
corneliusweiss.de/?p=53) was a bit more urgent

Hi devs,

i just worked out a proposal for history logging and a timemachine/
waybackmachine:
http://www.tine20.org/wiki/index.php/Developers/Egwbase/Timemachine

it would be great if some of you take a look at it and comment it, as
I want to start implementation of this stuff (for tine) between the
years.

Egwbase_Timemachine_Interface::getRecord() takes a string,
Egwbase_Timemachine_Model_ModificationLog::record_identifier is an
int. I may be mis-reading your diagram, but this seems inconsistent.
Will this work for a multipart key? I get those a lot.

Nathan Gray
nathan at goarctic dot com

---

Extra punctuation??? Ug!!!
Just say no! Punctuation marks do not gain significance if you
increase their number, they make you look silly. One is sufficient.

---

---

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

---

eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers

---

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

---

eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers

[Nathan_Gray-2]

Hi Nathan et all,

I just reworked the interfaces, and now they are hopefully consitent.
…
Here is the UML and description:
http://www.tine20.org/wiki/index.php/Developers/Egwbase/Timemachine

Egwbase_Record_Interface->getId() returns String,
Egwbase_Timemachine_Model_ModificationLog->record_identifier is of
type int. Maybe I don’t understand how this is supposed to tie
together, or maybe you haven’t had the chance to update
Egwbase_Record_Interface to reflect your datatype changes, but it
still seems inconsistant.

For Egwbase_Timemachine_ModificationLog::getModifications(…) is
_identifier supposed to be the
Egwbase_Timemachine_Model_ModificationLog identifier or the
record_identifier? I think record_identifier would be more useful, I
can’t see the use of identifier when you have the
getModification(_identifier) call.

If there are not other ideas, i’ll start coding tomorrow.

The multipart bit I mentioned is in pERP there are quite a few
documents with headers and child records. Sales orders with sales
order lines, invoices with invoice lines, shipments with shipment
lines. Also, I have multiple document types in the same application
(different tables of course). For example, in Accounts Receivable,
there are clients, invoices, receipts, etc.
I can link these using the current link system (which is great) by
using multi-part string keys for the same application: CLIENT:23,
INVOICE:213, and INVOICE:213:4 for linking directly to a line.

I suspect I could do history by using multipart modified_attribute,
such as CLIENT:status or SHIPMENT:expected_date similar to how I do
linking, but that would complicate everything. If record_identifier
is a String to match Egwbase_Record_Interface->getId(), then I could
put a multipart key like CLIENT:23 in there.
Stuff to keep in mind, but it probably affects me more than anyone else.

I had to postpone it a bit as work on child 2.0 (http://corneliusweiss.de/?p=53
)

Congratulations. We had a laugh at the release announcement.

Nathan Gray
nathan at goarctic dot com

---

Extra punctuation??? Ug!!!
Just say no! Punctuation marks do not gain significance if you
increase their number, they make you look silly. One is sufficient.

---

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

---

eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers

[Cornelius_Weiss]

Hi Nathan,

I just didn’t change the interfaces in Egwbase_Record_Interface to
int as identifier, that will come when i touch this class next time.

As for the multiple keys:
Maybe we should add a free form string field in the
Egwbase_Timemachine_Model_ModificationLog.
So each app can do it’s own magic here.

what do you think?

cu
conny

Hi Nathan et all,

I just reworked the interfaces, and now they are hopefully consitent.
…
Here is the UML and description:
http://www.tine20.org/wiki/index.php/Developers/Egwbase/Timemachine

Egwbase_Record_Interface->getId() returns String,
Egwbase_Timemachine_Model_ModificationLog->record_identifier is of
type int. Maybe I don’t understand how this is supposed to tie
together, or maybe you haven’t had the chance to update
Egwbase_Record_Interface to reflect your datatype changes, but it
still seems inconsistant.

For Egwbase_Timemachine_ModificationLog::getModifications(…) is
_identifier supposed to be the
Egwbase_Timemachine_Model_ModificationLog identifier or the
record_identifier? I think record_identifier would be more useful, I
can’t see the use of identifier when you have the
getModification(_identifier) call.

If there are not other ideas, i’ll start coding tomorrow.

The multipart bit I mentioned is in pERP there are quite a few
documents with headers and child records. Sales orders with sales
order lines, invoices with invoice lines, shipments with shipment
lines. Also, I have multiple document types in the same application
(different tables of course). For example, in Accounts Receivable,
there are clients, invoices, receipts, etc.
I can link these using the current link system (which is great) by
using multi-part string keys for the same application: CLIENT:23,
INVOICE:213, and INVOICE:213:4 for linking directly to a line.

I suspect I could do history by using multipart modified_attribute,
such as CLIENT:status or SHIPMENT:expected_date similar to how I do
linking, but that would complicate everything. If record_identifier
is a String to match Egwbase_Record_Interface->getId(), then I could
put a multipart key like CLIENT:23 in there.
Stuff to keep in mind, but it probably affects me more than anyone
else.

I had to postpone it a bit as work on child 2.0 (http://
corneliusweiss.de/?p=53
)

Congratulations. We had a laugh at the release announcement.

Nathan Gray
nathan at goarctic dot com

---

Extra punctuation??? Ug!!!
Just say no! Punctuation marks do not gain significance if you
increase their number, they make you look silly. One is sufficient.

---

---

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

---

eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers

---

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

---

eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers

[Nathan_Gray-2]

Hi Nathan,

I just didn’t change the interfaces in Egwbase_Record_Interface to
int as identifier, that will come when i touch this class next time.

As for the multiple keys:
Maybe we should add a free form string field in the
Egwbase_Timemachine_Model_ModificationLog.
So each app can do it’s own magic here.

what do you think?

record_type sounds good to me, so long as it’s filterable. You know
your design better than I, so if you think the cases I mentioned will
be satisfied by that, it should work.
I don’t know if you’ve thought about linking much yet, but I expect
that would be similar.

Nathan Gray
nathan at goarctic dot com

---

Extra punctuation??? Ug!!!
Just say no! Punctuation marks do not gain significance if you
increase their number, they make you look silly. One is sufficient.

---

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

---

eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers

[Cornelius_Weiss]

Hi Nathan,

i’m glad that there is a least one person reading my uml’s (even
before i announce them)

Thanks for the filter hint. I’m a bit brainded atm. as Child 2.0
costs lots of my nights

I allready finished the modificationLog implementation and inserted
it into the Tasks backend. As there is no Tasks frontend atm. i’ll
also include it into addressbook to have a test case

I’m not shure if i should implicitly solve concurrency problems, when
i can solve them with modificationLog (cause no confilct on field
basis) or If i should notify a web-browser client in any case.

For Sync devices i think i’ll allways solve conflicts silently to
avoid duplicates.

cu
Conny

Hi Nathan,

I just didn’t change the interfaces in Egwbase_Record_Interface to
int as identifier, that will come when i touch this class next time.

As for the multiple keys:
Maybe we should add a free form string field in the
Egwbase_Timemachine_Model_ModificationLog.
So each app can do it’s own magic here.

what do you think?

record_type sounds good to me, so long as it’s filterable. You know
your design better than I, so if you think the cases I mentioned will
be satisfied by that, it should work.
I don’t know if you’ve thought about linking much yet, but I expect
that would be similar.

Nathan Gray
nathan at goarctic dot com

---

Extra punctuation??? Ug!!!
Just say no! Punctuation marks do not gain significance if you
increase their number, they make you look silly. One is sufficient.

---

---

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

---

eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers

---

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

---

eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers