software installer onto a workstation?
Surely the earlier justification of things like ease of test string manipulation, and mainly
the authors productivity - are the significant arguments for this class of application?
John
Gabriel Paubert wrote:
On Thu, Oct 21, 2004 at 07:29:09AM +0000, Karel Kulhavy wrote:
On Thu, Oct 21, 2004 at 09:16:41AM +0200, Piotr Miarecki wrote:
I appreciate the fact these tools are in C and not in Perl or Python.Hi Stuart and All,
at first - I compromize everything if you only agree to work together on gEDA installer/setup, including Python. The only condition is that every step should carry a new functionality, not rewriting the existing code nor writing everything from scratch.
Majority of gEDA tools are one-person-projects. If a developer has no time for gEDA, temporarily or permanently, the tool is not maintained. A good example was setup and gmanager during past time. If you look into setup.cfg you see that it supported pcb, gwave, gtkwave and other tools - but in old, obsoleted versions. Because during past three years I had only few time for gEDA.
If you start a new installer project - this is what probably happens with your installer in the future as well. I propose that we cooperate. There are many features missing: updating support for non-gaf tools, porting to windows, creating a tool for creating setup.cfg, supporting APT/RPM(/PKG) databases, etc.
Answering your questions:
* setup HAS BEEN designed using glade, take a look into CVS: ~/glade/setup/setup.glade,
* setup does support ALL gEDA tools (now in obsoleted versions, it needs to be updated),
* setup is written in C, yes, because imo its a gaf standard, as well as guile should be used for scripting (imo it means also: not python, perl, etc.) - but this is a subject that should be separately discussed with Ales.
C is fast and there's no danger that some new version will come in which
something will behave differently.
Besides the run time bloat of Python in some cases. I'm fighting right
now with somebody who wants me to rewrite something which works and occupies (after compilation) 8kbytes (code+data in the executable) with a Python script which just after loading the libaries (before adding a
single line of code of my own) already takes up 6-7Mb, and he wants to use an XML parser on top of that (it's buzzword compliant for sure) which will take an additional unknown amount of RAM to run. BTW this has to run on a diskless machine with 16Mb of RAM...
Now I happen to hate glade. I'm writing a new app in C++ (my first C++
code so it has rough edges) and gtkmmm. It the first time I use Gtk and
I reserve my judgement for now, although I have the impression of being
fighting against the toolkit on each and every line of code (but I tried
glade and it was worse, I just can't stand a single line of automatically
generated code in my programs, and that's physical). For small projects like this one, I prefer Python+TkInter but in this case it can't provide the performance I need.
Regards,
Gabriel
