[Soc-coordination] Declarative Diversions - Report 5

[Sylvestre Ledru]

Hello Guys,

Thanks for your reports. However, it seems you forgot the 4-5 lines
summary about your whole GSoC (subject, results, etc).

Thanks
S

Le dimanche 21 août 2011 à 23:09 +0100, Sam Dunne a écrit :
> DECLARATIVE DIVERSIONS
> ======================
> 
> REPORT 5 - 21st August 2011
> -------------------------
> This is the fifth and last report for declarative diversions for Debian as part
> of Google Summer of Code.
> 
> 
> WHAT HAS HAPPENED SINCE LAST TIME
> ---------------------------------
> The last two weeks have been a rush. The amount of work has piled up
> quite a bit. This happened for a number of reasons.
> 
> 1) There was a lot of refactoring that was done in "lib/dpkg/parse.c"
> that was needed for the project to be successful. guillem did all of
> this for us which was a massive help
> 2) I was sick last week and had a hard time concentrating on code so
> progress was slow
> 
> However the last three days has been extremely productive. Myself and
> vorlon have been working well to fix issues with the code. We have
> replicated a lot of code contained within parse.c
> to keep the code as closely related to the existing codebase as possible.
> 
> Our general program flowchart (as of now) goes like this:
> divert_parse_ci()		Called once for each file
>  |
> V
> parse_stanza() 		Called once for each stanza. parse_EOF() tells us
> there are no more
>  |
> V
> diversion_parse_field() 	Called once for each field in the stanza
> 
> We had to remove the ability to place comments within the diversions
> file. Because as of right now it will break the parser.
> This evening (21st August 2011) the code builds successfully. The
> parser works and creates a linked list successfully
> 
> Unfortunatley the code still has a bit of doing and will not be
> completed in time. I plan on staying around either way and the project
> will
> be finished off. There is a todo list ;)
> 
> 
> PLANS FOR THE COMING FUTURE
> ---------------------------
> Finish off everything on the todo list:
> * If a field occurs more than once in a stance throw an error
> * Fix diversion_ctor to not use static arrays for our strings
> * Handle the case of a declarative diversion with no "To:" specified,
> by filling in the default value based on the from field
> * Check if there is a difference in old and new diversions "To" field
> when calculating diversions
> * Function to free the lists we allocate
> * Integrate with processarc.c
> * Drop testing main() function
> * Makefile.am: add divert.c and divert.h into the list of dpkg sources
> 
> Stay on with Debian with the aim to become a Debian developer
> 
> 
> WHAT I FOUND HARD
> -----------------
> Throughout the project the whole concept of data structures has been
> challenging as it was completely new to me. However vorlon has
> explained a lot of what I need to know extremely well and I am
> confident that I know far far more than I did 3 months ago on the
> topic
> 
> 
> WHAT HAS GONE WELL
> ------------------
> In short, the Debian family :)
> 
> 
> CONTACT ME
> ----------
> Anyone who wishes to discuss the project with me can either mail me through
> the mailing list, catch me on IRC (SCD[Eire]) or email me at
> sam.dunne101 at gmail.com
> 
> My blog:    http://blog.sam-dunne.com
> My Github:  https://github.com/samdunne/dpkg-diversions
> 
> My code is located in '/lib/dpkg' in the 'testing' branch of my code. The files
> are 'divert.c' and 'divert.h'
> 
> *Note* - You will have to also checkout '/lib/dpkg/div_test/' folder for my code
> to work
> 
> 
> A final thanks as a GSoC Student :),
> ~Sam
> 
> 
>