Get rid of the libcurl-dev dependency
rhornig opened this issue · 2 comments
The dependency from libcurl is rather uncomfortable, because on windows OMNeT++ does not com with the libcurl-dev so the user has to set up the package manually. libcurl is used by the model to actually reach out to the internet and pull the necessary data. This prevents repeatable experiments as the data is always refreshed.
I would rather suggest to drop all the code that is dealing with the download. Instead the model should utilize the files that are already downloaded from the internet (i.e. assume that they are present). If you need fresh files those can be downloaded using a shell script (probably before starting the app).
The wget or curl commads can be utilized for this and they are readily available on all platforms.
Rudolf,
I agree with your suggestion about (libcurl) and probably any 'live'
internet accessing it is doing -- I'll try to work soon -- in fact it
might be sooner than soon, as I'm getting tired of using the configurator
tool (which has bugs as I've discovered) only for it to download a file
off Celestrak and modify the omnet.ini file setting ground sites and
atmospheric parameters.
I do like what I see in the OS3 software, but there is a strong need for
simple examples to get a new user/developer up and running and
comfortable with what they are looking at. Right now, the 'main'
example requires way too much knowledge (and downloaded 'live' data) to run.
I just tagged the 'original' code as version 1.0 (git v1.0 tag). As I
update to more current inet releases, I'll tag at logical points. Really
want to get to inet 2.99 as soon as possible, but I might tag a release
that uses 2.5 first.
Doug
On 02/11/2015 02:59 PM, Rudolf Hornig wrote:
The dependency from libcurl is rather uncomfortable, because on
windows OMNeT++ does not com with the libcurl-dev so the user has to
set up the package manually. libcurl is used by the model to actually
reach out to the internet and pull the necessary data. This prevents
repeatable experiments as the data is always refreshed.I would rather suggest to drop all the code that is dealing with the
download. Instead the model should utilize the files that are already
downloaded from the internet (i.e. assume that they are present). If
you need fresh files those can be downloaded using a shell script
(probably before starting the app).The wget or curl commads can be utilized for this and they are readily
available on all platforms.—
Reply to this email directly or view it on GitHub
#1.
Reviewing code now that touches "WebServiceControl" which grabs data
from websites (it uses libcurl) -- interesting code, but also makes
things a bit complicated to run -- easy to remove. Just create a file
with TLE data in it and be done with it.
On 02/11/2015 02:59 PM, Rudolf Hornig wrote:
The dependency from libcurl is rather uncomfortable, because on
windows OMNeT++ does not com with the libcurl-dev so the user has to
set up the package manually. libcurl is used by the model to actually
reach out to the internet and pull the necessary data. This prevents
repeatable experiments as the data is always refreshed.I would rather suggest to drop all the code that is dealing with the
download. Instead the model should utilize the files that are already
downloaded from the internet (i.e. assume that they are present). If
you need fresh files those can be downloaded using a shell script
(probably before starting the app).The wget or curl commads can be utilized for this and they are readily
available on all platforms.—
Reply to this email directly or view it on GitHub
#1.