GSoC 2008 Proposal: “Packagable GT” (Organization: The Globus Alliance)

This is the text of my proposal for the Google Summer of Code 2008 “Packagable GT”. The Organization is The Globus Alliance. All comments are welcome.

“Packagable GT”

1 Abstract (2500 characters limit)

1.1 Proposal Name

“Packagable GT” (Organization: The Globus Alliance)

1.2 Problem

Lack of a Globus Toolkit (GT) source building process that allows eventual compatibility with standard GNU/Linux packaging systems.

1.3 Purpose

To ease GT source’s eventual packaging into standard GNU/Linux distributions

1.4 General Objective:

• To adapt the current GT source building process in order to enable its eventual packaging into standard GNU/Linux distributions

1.5 Deliverables

For the Mid-term Evaluation:

• A complete analysis of the GT building process, with details on how GPT specific actions could be modularized.

For the Final Evaluation:

• Modified GT building process, easier to package into a Deb/RPM-based GNU/Linux distribution

• (Optional) GT packages for mainstream GNU/Linux distributions (especially Debian/Ubuntu Debs and SuSE/Fedora/Mandriva RPMs)

1.6 Web Site

http://allentiak.wordpress.com

2 Detailed Proposal (7500 characters limit)

2.1 Proposal Name

“Packagable GT”

2.2 Problem

Lack of a Globus Toolkit (GT) source building process that allows eventual compatibility with standard GNU/Linux packaging systems.

2.3 Purpose

To ease GT source’s eventual packaging into standard GNU/Linux distributions

2.4 General Objective:

• To adapt the current GT source building process in order to enable its eventual packaging into standard GNU/Linux distributions

Specific Objectives

• To classify each step of the current GT source building process into one (or more) of three class of functions (binary building, configuration files generation and installation of the two previous elements) and analyze the inter-dependencies between all of them.

• For each source building process step, make the GPT work in a modular way (to separate its functions into the three classes defined above).

2.5 Deliverables

For the Mid-term Evaluation:

• A complete analysis of the GT building process, with details on how GPT specific actions could be modularized.

For the Final Evaluation:

• Modified GT building process, easier to package into a Deb/RPM-based GNU/Linux distribution

• Documentation on the new building process

• (Optional) GT packages for mainstream GNU/Linux distributions: at least one using each packaging format (Deb and RPM).

2.6 Benefit for the GT Community

To allow the eventual integration of GT into popular GNU/Linux distributions, given that independent binary and configuration files will be easier to build and to maintain.

2.7 Motivation

GT uses its own packaging system, GPT. While this packaging system might be appropriate for GT, it is rather too GT-specific. There is a mixture between the building, configuration and installation sections of the entire GT components (both binaries and configuration files)’ creation process[-1].

Such a situation makes impossible to package the GT following any distribution packaging policy in a way that allows coherent configuration of the installed binaries. Thus not allowing the growing of the GT’s popularity.

In order to allow the GT to be packaged into any distribution, a modified GPT building process that separates the three parts defined above is needed. Given that GNU/Linux is the most popular choice when deploying a grid solution, the packaging formats to consider will be Deb and RPM.

However, to achieve 100% compatibility with any of the previously mentioned packaging standards is not easy. Both RPM and Deb have their specific configuration and dependencies aspects; and the GT building process is complex.
That is why a RPM/Deb 100% compatible GT package is an **optional** deliverable of this proposal. In case this is not possible, the resulting modified GPT-based process should allow (or, at least, ease) an eventual RPM/Deb packaging.

2.8 Activities

This is how it is intended to proceed.

1. Research on how GPT works.

2. Analysis of the details of the current GT building process (GPT-based) in order to find non-isolated tasks (building, configuration and installation). This step is crucial, given the size and complexity of the GT source code.

3. Research on how Deb and RPM work.

4. Mocking up of possible ways to modify the current GT building process in order to isolate the previously defined types of tasks.

5. Modification of the current GT building process (GPT-based) so all tasks get isolated.

6. Proof of Concept:

(a) Building of new GT packages based on the modified GPT-based building process.

(b) Ideally, new GT Deb and/or RPM packages.

2.9 Considerations

• This project will be the student’s main activity during the considered period.

• Weekly status reports will be made.

• The communication with the rest of the GT Community will be fluent.

2.10 About the Student (Section written in first person)

My name is Leandro Doctors; I am 26 years old. I live in San Juan (Argentina), a small town located 1000 km to the west of Buenos Aires, not too far from the border with Chile.

Contact Info

• Name: Leandro Doctors

• E-Mail: ldoctors-at-gusila.org.ar

• XMMP (Jabber): allentiak-at-lugmen.org.ar; ldoctors-at-gmail.com

• Web Site: http://allentiak.wordpress.com

• Timezone: UTC -3 (minus three)

Educational and Technical Background

I am a student at the Faculty of Exact, Physical and Natural Sciences (FCEFyN)[0] of the National University of San Juan (UNSJ)[1]. I graduated one month ago from my first major (Information Systems) and I started my second one (Computer Science). My IS thesis[2] (in Spanish) describes the potential application of grid technology to the FCEFyN campus and includes a working prototype as a proof of concept. I love languages: besides my native Spanish, I speak fluent English (B2 CEFRL[3] level) and French (B1) and some Portuguese (A1) and German (A1).

You will find a résumé of my skills and abilities in my weblog[2].

Free Software Background

I am a GNU/Linux user since 1999 and I have evolved into a free software activist in 2002. I am co-founder of a small Free Software Users Group: GUSILA[3]. Although it is currently not so active right now, GUSILA has contributed to highlight Free Software as a Subject of discussion into the City of San Juan. Specially worth mentioning is the organization of the local editions of the FLISoL (Latin American Free Software Install Fest) in San Juan, in 2005, 2006[4] and 2007[5].
Also, I have contributed PHP code to the Pandora Free Monitoring System[6].

Grid Computing Background

I have been investigating about grid computing for about one year, in order to write my grade thesis for my Information Systems major. It describes the potential application of grid technology to the campus of my Faculty (the FCEFyN). While researching, I begun to like grid computing to the point it has become my main field of research since. Currently, I’m working towards the establishment of a grid and cluster computing research group at the Institute of Informatics (IdeI) of the FCEFyN, in collaboration with members of the Argentinian Grid Community.

References

Regarding my modeling abilities, you can ask Mrs. María Inés Lund[7], one of my professors at the UNSJ. Regarding my research group project, you can ask Javier Echaiz[8], from the National University of the South (UNS) and Sergio Zapata[9], the Director of the IdeI. Regarding my contribution to the Pandora FMS project, you can ask Sancho Lerena[10]. If you need other references, just let me know.

I think that my knowledge of Analysis and Design Languages (such as UML) along with A&D techniques (such as the UP) plus some knowledge about Knowledge Management can help me be the right man for the job.

[-1] Tibor Kalman (instant-grid.de Project) tibor.kalman-at-gwdg.de: Private Conversation.

[0] FCEFyN: http://www.fcefn.unsj-cuim.edu.ar/

[1] UNSJ: http://www.unsj.edu.ar

[2] Web log / Web site http://allentiak.wordpress.com

[3] GUSILA: http:/www.gusila.org.ar

[4] FLISoL 2006@San Juan: http://installfest.info/FLISOL2006/Argentina/SanJuan

[5] FLISoL 2006@Rivadavia,San Juan: http://installfest.info/FLISOL2007/Argentina/Rivadavia

[6] http://pandora.sourceforge.net/en/index.php?sec=about

[7] María Inés Lund: mlund-at-iinfo.unsj.edu.ar

[8] Javier Echaiz: jechaiz-at-cs.uns.edu.ar

[9] Sergio Zapata: szapata-at-iinfo.unsj.edu.ar

[10] Sancho Lerena: slerena-at-gmail.com