Closed GoogleCodeExporter closed 8 years ago
When C++11 becomes more fully accepted, I will remove the boost dependency and
switch to std::shared_ptr, etc. In the meantime, however, the shared pointers
are just too useful.
Out of curiosity, why can't you include the boost code in your library? The
number of headers bloat the size of the binary, so it shouldn't be a problem
for embedded systems.
Original comment by jbe...@gmail.com
on 11 Apr 2013 at 4:15
shared_ptr is great - but if you use the tr1 version, it's just a single file -
why not throw that in and be done with it, I'm pretty sure it doesn't use any
C++11 features?
The reason I'm loathe to throw in the Boost files is that 1200 files is many
times as many files are in this project already - and the functionality I'm
gaining is a tiny boost in my personal satisfaction for a tool where no one
else sees the interface (and, I believe, greater reliability in future - Yaml
is definitely a better way to go).
This application is at least partly for academic purposes and I'm just not
willing to stand up and explain that "These 1200 files are there to support
these 30 lines of code in this utility" - particularly when if I stick with
0.3.0 I don't have to do any of this, just drag a few files into a project
directory.
I wasn't looking for trouble - I simply started with doing what was natural,
downloading the whole Boost library. At a certain point, I realized that we
were talking almost half a gig when expanded and that I'd blow through my
"disk" space.
I am allowing some external dependencies - well, one, numpy - so it's not
unreasonable to ask people to install something - but I can't expect people to
use half a gig of disk space on a machine where the typical "disk" is 4G.
Then I went to my main desktop and started to figure out how to extract a
subset of Boost. After some work, I had a list of 1200+ files and realized
that I should have just installed 0.3.0, the version I'm using everywhere else,
an hour before...
I do love this library, don't get me wrong.
I was only interested in 0.5.0 because I thought it'd have a fix to Issue 148,
but I have a workaround for that bug anyway (I read and write my own ---
markers before yaml-cpp ever sees them). 0.3.0 works fine for me, I have a
separate code base that makes heavy use of it, and there's no incentive to move
forward!
Original comment by tom.ritc...@gmail.com
on 11 Apr 2013 at 7:52
OK, fair enough - I will still support 0.3.x with bug fixes. In particular,
when I fix Issue 148, I will fix it for both branches.
(By the way, I realized I'm also using boost::iterator_facade and friends,
which would make it more difficult to remove the dependency. I probably still
will, since I understand the value of minimizing dependencies, but it will just
be more difficult.)
Original comment by jbe...@gmail.com
on 12 Apr 2013 at 3:38
I was excited to use yaml-cpp for my little project but when I saw it had a
Boost dependency... that ended my excitement. I'm looking for another solution.
Hopefully, some day you'll move away from the Boost dependency.
Original comment by colbymin...@gmail.com
on 15 Oct 2014 at 6:10
+1, the boost dependency was a show stopper for me unfortunately.
Original comment by flo...@gmail.com
on 23 Mar 2015 at 11:12
I can't use your project due to the boost dependency. Boost is just too much to
deal with. It would have been especially nice if you had actually listed boost
as a dependency in the documentation of your project so I could have moved on
without wasting my time trying to build your project.
Original comment by terreyw...@gmail.com
on 21 Apr 2015 at 2:46
Original issue reported on code.google.com by
tom.ritc...@gmail.com
on 11 Apr 2013 at 1:32