After new achitechture's been designed as a bluescript, more problems emerge.

21:23 2006-9-5
trouble list:
1. RenderDevice's CDC member can't be a member, for CDC* is passed in dynamically?
   but being member is more suitable for Opengl Device.. so it's a conflict, and it's emergent.
2. Decorator's responsiblity and that of MidiScore are messed up and obscure. they both access notes frequently. I wonder whether Decorator should be eliminated.

3. some implementation trouble: how to locate a note in score by cursor position, where to store the posTable if any? how to search the note more effectively?
how to insert new notes after an existing note changed?
... more implement details.

4. should RenderTools (pen, brush, fonts) be global or Tool's member, or Ploter's or RenderDevice? Should be paid more consideration, but if don't go further on implementation, these problem can't be found.. that's awful, really.

I have to say..

This is the biggest and hardest project I've ever had!

and now I'm leaving, and back to school.

and the challenging is grow bigger, and I gotta prepared!

Fareware, all peaple here, gived me a nice time in Beijing.

Gotta Move Out!

before sleeping, I think about some main problem getting in the way. And I got a hobby of thinking and reading before sleeping after a whole day's work.

but today is efficient , like some other day I came into an idea all in a sudden!

and smoothly , I take out my pen, scratching on the paper what I got.

GENERAL IDEA IS:

      1. conclusion on GROUP notes:

                   - 1. Group end when any condition below met:  rest. 

Didn't do much on Core Algorithm.

just pay attention to the drawing method.

get more familiar with CDC and get rid of the annoying glittering problem when draging the image.

tried to turn to GDI+, but now it will take a big effort to change current framework to fit GDI+ properly, so concern it later.

and the holiday is ending....

being pulzzling and confusing and excited...

I found a big mistake.

division in the midi file header always represent the timecount of QUARTER note, not that of a BEAT!

I really mistook the concept though after reading couple of documents,  but it's ok I find the problem in time

new task... more tasks...

new bugs... more bugs...

Kind of missing OpenGL!...

After all these days messing up with GDI(in monster MFC).......

If I use OpenGL, many things will be just as beautiful as it rawly is, and MOSTLY efficient!

A solution to solve the print preview using Opengl is : (my own opinion)

        Before switching to print mode, instantly get a screenshot through BitBlt method of pixels that Opengl draws, and draw the temporarily pixel block or say, a texture in buffer to the print DC. That' s the general idea,

wait to be implemented..........

before 8.23 some items already fixed:

item 2.  3.  6.

leave: 1.  4.  5.  7

TODAY:

item 4 half-done, greatly enhanced beam display method! I'll show some code below.

newly added: (Better done before 8.26) 

                     8.   timemark now inconsistent, ie 3/4 or 3/8 will encounter error.

                     9.   DIVIDE SCORE INTO 2 PARARREL PARTS , and this is come up with me yesterday night.

                   10.  more reasonable grouping rules, now even rest can be grouped in.

Requirement Modified....

1. Fully iteration: divide staff into two logical parts -- high region and low region...  then deem them as two seperate track , managing their respect notes and rests but still draw on a whole staff.  It'll be quite a reconstruction.

2. all notes'bar up in high region , all notes'bar down in low region

3. beam 's direction same as notes by region

...

Besides note's grouping module, some trival but important problem to be fixed.

TO DO LIST:

1. note's lengthTotal and lengthSounded field confusion. To treat by note type.

2. make pitch range absolute.

3. add custom setting range.

4. beam modification and enhancement

5. concerning note that covers notes.

6. bar color modification, to be more common.

7. concerning note cross two adjacent measures.

... (to be continued)

Reinforcement

Weekend And Take A Break!

Today went to the Greate Wall. Climb SO HIGH.

Other Times Eating, Joggling, And In a Car.

And after that, backing to work...

I know why the parser corrupts when read some strange file.

'cause I don't cover some very rare midi messages -- the Channerl Mode Message(found from an elaborated document).  Most of file doesn't have them.

and the message has a same form to one of the Channel Sound Message: Set Volume 0xbf,  what can be deadly confusing.

Ok , add the msg to current parsing system, it will be more robust

Try a simple way to merge notes.

some excerpt from <<The AB Guide To Music Theory>> about grouping note:

Note Grouping:
1. The standard pracitse is to avoid ties where possible.
   Bear in mind that ties join notes of the same picth.

Beam Add Rule:

1. Notes shorter than quavers are beamed together in beats

2. demisemiquavers(32th) may also be grouped in half-beats

3. other ways to write : add a single or double short beam between 2 half-beats groups.

不停刷新就狂加访问量

这个做的不合理啊， 希望加强一下， 看看LBS这方面做的就很好

另外浏览器一关闭就又要重新登陆， 很不人性啊， 登陆信息应该驻留一段时间的啊， 应该有个选项吧

刚刚又发现当发表评论后出现“文章发表成功”.... 如果这是在测试， 就赶快修复bug吧