[BitPim-devel] Re: [bitpim-cvs-checkins] bitpim/help screen-editin gcalendarevent1.png,NONE,1.1 screen-editingcalendarevent2.png,NONE,1.1 screen-editingcalendarevent3.png,NONE,1.1 screen-editingcalendarevent 4.png,NONE,1.1 screen-editingcalendarevent5.pn

[d***@netzero.com]

All those files need to be updated for the help to work correctly when >not running in a developer tree.

I figure those files would be generated as part of the build and that's why I didn't commit them. btw, what's the process of generating new help sections (those that don't yet have Help IDs?) I'm thinking specifically of the Calendar Import Dialogs.

-Joe Pham



______________________________________________________________________
Speed up your surfing with NetZero HiSpeed.
Now includes pop-up blocker!
Only $14.95/month -visit http://www.netzero.com/surf to sign up today!

[Roger Binns]

Post by d***@netzero.com

All those files need to be updated for the help to work correctly when
not running in a developer tree.

I figure those files would be generated as part of the build and that's
why I didn't commit them.

The wxh file is the master project file and others are generated from
it in various ways. They all have to line up right.

Post by d***@netzero.com
btw, what's the process of generating new help sections (those that don't
yet have Help IDs?) I'm thinking specifically of the Calendar Import Dialogs.

For new files you need to tell me where you want them in the
hierarchy and what the names are. If you need a new id then
also tell me what it is and what file it is associated with.
I do the editing in Help Blocks which updates the project
and associated files, and also run the help generator.

Unfortunately HelpBlocks isn't very good as a collaborative
tool, but it does the best job for the price.

BTW there are two types of ids. The numerical ones are only used in
the web version to link pages into the right place in the frame of the
content structure. Those are automatically updated by the help
build system. You can just put in 0 and it will be updated in the
next help build.

The string ids are used when you need to make a help button in
the UI go to the corresponding page in the help. That involves
updating page properties in the Help Blocks UI.

Roger

[d***@netzero.com]

For new files you need to tell me where you want them in the hierarchy >and what the names are

Please add the following files:

Reference/Dialog boxes/Calendar Import: dialog-calendarimport.htm
Reference/Howtos/Import Calendar Data: howtos-calendarimport.htm

If you need a new id then also tell me what it is and what file it is >associated with

Please add the following IDs:

ID_DLG_CALENDAR_IMPORT="dialog-calendarimport.htm"
ID_HOWTOS_CALENDAR_IMPORT="howtos-calendarimport.htm"

Thanks,

-Joe Pham


______________________________________________________________________
Speed up your surfing with NetZero HiSpeed.
Now includes pop-up blocker!
Only $14.95/month -visit http://www.netzero.com/surf to sign up today!

[Roger Binns]

Ok, I've add the two new document files, the image files, the ids and
rebuilt and re-indexed everything.

Roger

[Vic Heintz]

This must have something to do with the scoring system for matching
names. It seems that if the imported vCard "FN" field does not exactly
match the BitPim "Full" name field, then other fields will likely take
higher precedence in deciding the match. This leads to some crazy
results.

I had entered phone numbers for a relative and her two next door
neighbors into the phone and then retrieved them into bitpim's database
with get phonebook. All had the same area code and similar if not the
same exchange. As with most of my relatives I edited the "Full" name
field to be a nickname because that is what I want to appear on the
phone. (eg First=John, Last=Smith, Full=Smitty; [names changed from
actual]) The neighbors were as might be expected (ie First=Fred,
Last=Johnson, Full= Fred Johnson and First=Mary, Last=Jones, Full=Mary
Jones)

Now I added addresses to John Smith, Fred Johnson, and Mary Jones in
Mac OS AddressBook and exported a vCard file. Naturally, as neighbors,
they all had similar addresses with same street, city, state and zip
code. I believe the first address, say Fred Johnson, imported
correctly because there was an exact "FN"="Full" match. I don't
remember for sure about the second (ie Mary Jones.) But BitPim
persisted in wanting to add John Smith's address as a second address in
Fred Johnson's record. This is in spite of there being an exact first
and last name match for John Smith even though full name "Smitty" did
not match "John Smith" (as it also did not match "Fred Johnson") I
presume this is due to the good match for area code, address, and zip
code taking precedence. (There was not yet an address and zip code in
Smitty's bitpim record to match on.)

I don't know what the fix to BitPim would be other than to give a
higher score to matching first and last name when there is a poor full
name match. My work-around was to temporarily edit the full name field
to be an exact match, do the import, and replace it with my preferred
familiar name.

Vic

[Roger Binns]

Post by Vic Heintz
I don't know what the fix to BitPim would be other than to give a
higher score to matching first and last name when there is a poor full
name match.

The fix is something I have been requesting for a long time. Produce two
vCard (or whatever format you want) files to be used as a test suite.
The idea is that you clear all entries and add the first. Then import
the second which should be deliberately contrived to flex and tune
the matching rules.

The scoring and weighting is currently rubbish. I never had any data
to do good testing against and no time to make the data. Consequently
the code uses what amounts to random constants.

Roger

[Vic Heintz]

Post by Roger Binns

Post by Vic Heintz
I don't know what the fix to BitPim would be other than to give a
higher score to matching first and last name when there is a poor
full name match.

The fix is something I have been requesting for a long time. Produce
two
vCard (or whatever format you want) files to be used as a test suite.
The idea is that you clear all entries and add the first. Then import
the second which should be deliberately contrived to flex and tune
the matching rules.
The scoring and weighting is currently rubbish. I never had any data
to do good testing against and no time to make the data. Consequently
the code uses what amounts to random constants.

I have been unable to duplicate my problem outside of my existing
BitPim database using only vcard files. Even when I export my entire
phonebook as a vCard file and import it into a blank database, the
vCards created by export from AddressBook seem to import OK. But my
existing database still has reproducible problems importing vCards.

I made a copy of my database and stripped out all but 2 records (my
relative and the next-door neighbor). I have a vCard with just my
relative's info exported from Mac OS AddressBook. BitPim still wants to
merge my relative's info into the neighbor's record. I would be willing
to privately send these files to someone willing to look at this and
figure out what is going on.

Vic

[Roger Binns]

But my existing database still has reproducible problems importing vCards.

Maybe your serials are confusing things. Turn on the developer console
and then run this code:

import common
open("foo.idx", "wt").write(common.prettyprintdict(db.getmajordictvalues("phonebook")))

Then have a look in foo.idx

Roger