Re: Student proposal deadline.

From: Ersin Akinci <ersin.akinci_at_gmail.com>
Date: Sat Apr 09 2011 - 08:43:33 CEST

Hi all,

It seems like the mistake is on me; actually, I think that Martin was
providing a great heads up by saying that the deadline was at midnight
rather than the actual deadline, which was in fact 11 AM PST. I got
confused, however, and thought that this meant midnight April 9th,
i.e., pitch black outside at the very end of April 8th. At least,
this is the standard usage of the term "midnight" where I am from, but
I've looked it up the term and I see that it's ambiguous. It refers
generally to the transition between night and day.

I sent a brief e-mail to Carol explaining the situation and CC'ed her
on this. I sincerely hope that this doesn't disqualify me from GSoC,
as I was really looking forward to the opportunity to work with
AbiWord again this year, but I'm bracing for the worst. At any rate,
here's my application. Best of luck to all the other applicants!

Best,
Ersin

Name: Ersin Akinci
E-mail: ersin.akinci@gmail.com
Project title: Improving performance by implementing a background
thread framework in AbiWord
Synopsis:
My project would impelment a background thread framework through which
preexisting and new functions in AbiWord could be re/implemented in an
intelligent, transactional, and transparent way that would shift work
to idle cycles.

Benefits:
Editing large documents would become feasible and there would be
significant performance improvements on all platforms, especially
embedded and low-power platforms where AbiWord is particularly popular
(or the only practical choice). Also, a general framework would allow
many future functions, not just those tied to breakSection or
rendering, to be impelmented in the background. This will be
particularly important as AbiWord takes on new filters and new
platforms, and especially now that multicore processing has become
popular.

Deliverables:
At the end of my project, AbiWord will be able to edit a 1000 page
text document with ease (i.e., without any delays when pressing
delete, backspace, enter, entering a character or a new line, etc.) A
new background thread framework will allow new and preexisting methods
to be re/implemented in an easy and transactional way that would allow
said methods to work when the processor was idle or to prioritize
which parts of the document/layout to update in the background to
improve efficiency.

Project details:
AbiWord is a modern word processor that deals with small documents
efficiently and comfortably. However, performance significantly
degrades with large documents (especially over 1000 pages), even on
modern high-powered systems. Although most of AbiWord's core
algorithims and data structures are sound and very efficient, they
have difficulty processing large documents because AbiWord lacks a
background thread framework for putting the relevant processes in the
background. Hence, when the "Enter" key is pressed, a method iterates
through the entire document to find the place where a new "container"
object should be inserted and freezes the word processor in the
meanwhile, whereas a more efficient implementation would move this
processing to the background since most users who are typing "enter"
are probably in the middle of typing a paragraph, etc., and do not
actually need to access other pages.

My project would therefore create a background thread framework
through which such critical functions could be processed during idle
cycles when AbiWord is not being used and that would lock the relevant
parts of the program while data is being processed (e.g., when user
scrolls down after pressing enter, pages that had not been reformatted
would be displayed as blank). Moreover, I would reimplement one
specific method, breakSection, using this framework, which would
complete my work from last year's GSoC to allow the practical
manipluation of large documents.

Project schedule:
One month before the project starts, I will look into background
thread impelmentations, how these may vary according to each of
AbiWord's platforms, and how these may be impelmented in AbiWord.
This will allow me to focus my full efforts on creating such a
framework for AbiWord. I have no precommitments for this summer and
am ready to work 40 hrs./wk. on my project.

Bio:
I am a doctoral candidate in the history department at the University
of Illinois at Urbana-Champaign studying the history of information
technology and science, and I have worked with AbiWord on a number of
occasions, most recently in GSoC 2010 to implement a new rendering
algorithm to improve performance with large documents. Through my
efforts, I have become intimately familiar with AbiWord's rendering
engine and would be well-positioned to tackle the job of implementing
a background thread framework, especially one that would allow the
all-important breakSection method to work when AbiWord is idle. My
current project proposal is strictly limited to improving performance,
as opposed to my previous proposal, which attempted to make AbiWord
page size and page position-agnostic and which proved to be a near
impossible task due to deeply embedded assumptions about document
structure. Givent that I eventually ported my performance-enhancing
code from GSoC 2010 to trunk after dropping the parts pertaining to
page size, I am confident that with a purely performance-oriented
proposal this year, I will be able to deliver a finished project to be
integrated with trunk immediately by the deadline.

On Fri, Apr 8, 2011 at 6:30 AM, Martin Sevior <msevior@gmail.com> wrote:
>
> Hi everyone,
>
> Now according to the Google Summer of Code timeline, the deadline for
> student proposals is the midnight of April 8th as measured in
> California, USA.
>
> After this time the AbiWord mentors will assess the proposals and may
> ask for for further information for you. If you haven't submitted a
> proposal by midnight April 8th you won't get into the GSoC program for
> 2011.
>
>
> Thanks to everyone who have submitted their applications for GSoC with
> AbiWord this year. It's fantastic to see so many high quality
> applications.
>
> Cheers
>
> Martin
>

-- 
Ersin Y. Akinci -- ersinakinci.com
What Digital Revolution? -- www.whatdigitalrevolution.com
Thinking critically about digital worlds.
Received on Sat Apr 9 08:44:06 2011

This archive was generated by hypermail 2.1.8 : Sat Apr 09 2011 - 08:44:06 CEST