The tl;dr; is that all projects are going well. All mentors have confirmed 
satisfaction with progress.


*Boost.Compute, Jakub Szuppe*

Detailed report: 
https://docs.google.com/document/d/1OwAqPjWBLF3B1YtN0_e87v1H_r41DE728kJk9mR7q3w/edit?usp=sharing

I started at the end of April with fixing find_end() that was randomly 
failing with SIGABRT error. Then I did various changes that allowed running 
tests on OSX on Travis CI. Later I added dedicated Travis build for 
Coveralls and builds in which the latest Boost libraries (1.61) are used. 

Next issue was resolving type-safety issues when copying to or from a host 
memory. It turned out to be the biggest task so far in terms of added lines 
of code (over 2300 LOC). Since copying is a crucial operation I worked on 
getting the best performance and added a lot of new tests to make sure it 
works in every possible case. I introduced three methods of copying data 
and which one will be selected to use depends on both the size of the data 
and the type of the device. 

The last task of the first half of BSoC’16 was implementing four comparison 
sort algorithms: sort(), sort_by_key(), stable_sort() and 
stable_sort_by_key(). They are all based on the same stable counting merge 
algorithm. Additionally, for non-stable sorts there is a block sort step 
(i.e. sorting subarrays) based on bitonic sorter which uses local memory 
and therefore is fast and improves the overall performance.

All scheduled tasks were successfully completed.



*Boost.Document, Anurag Ghosh*

1. Week1 : Testing improvements.

Progress: Code coverage is currently 89% (as of writing, it was 86% then),
would try to improve to 90+. Appveyor is found to be unsuitable for
Windows testing for the project as the dependencies (Libreoffice or MS
Office is not on their set of provided software).

I changed the order of work a bit so that things which flow well with each
other are together.

2. Week2-3: Comparisons

Progress: Done, comparisons have been added and tests have been added.

3. Week4-5: Cell as a reference.

Progress: Partially done. swap has been implemented, added a couple of
copy constructors. Tests have been written employing basic STL to check if
cells function as a reference. However, std::sort doesn't call std::swap
every time which has created a design issue. Antony has asked me to move
forward while we can find a workaround.

4. Week6-7: Cell formatting options.

Progress: Identified the MS Office and Libreoffice API calls. Was trying
to see which options are cheap and essential and can be standardized
across both the API's. Will write and push the code by the end of this
week. Also, can we directly change styles for ranges or not in Libreoffice
(in MSOffice we can) is currently what I'm looking at. Another issue is
automatically testing the code (without someone checking the saved file to
analyze behavior).

I have been a little behind (I had a research paper deadline, this is my
coursework) but I think I can catch up and finish up in time. Also, I
think week11-12 objectives would need to changed to something else, given
the preliminary findings in week1 regarding Appveyor. Other ways can be
explored


*Boost.Http, Vinícius dos Santos Oliveira*

Behind schedule, mentor apologies to SC for his tardy feedback to student 
which has delayed progress of student and was not fault of student, is 
confident student will make up ground now.


-- 
You received this message because you are subscribed to the Google Groups 
"Boost Steering Committee" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to