[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: HTTPAPI vs ISockets
Thank you for your kind words. I'd like to give my feelings on what
Cozzi said, if that's okay...
I wrote HTTPAPI largely for myself and one other developer to use. I
didn't really design/write it with the goal of it being a publicly
available software product. I later thought to myself "hey, it's useful
... maybe others would like a copy" so I put an open source license on
it and put it on my web site.
My attitude was "if this is useful, feel free to take it." I kinda
thought that was a nice thing to do.
Apparently not. Apparently if I want to give away source code, I'm being
downright rude, unless I offer to provide full support to the world on
my own time, for free.
Incidentally, for the most part, I actually *do* provide full support
for free. I just don't make any promises that I will always be there to
The library issue that Cozzi complains about is an example. As soon as
someone told me that it'd be better to remove the hard-coded libraries,
I did. Note that Cozzi never even mentioned the library issue to me. He
never gave me the option to fix it for him -- he, apparently, just
decided that I'd take the 'you've got the source, fix it yourself'
stance -- but since he didn't ask me (he didn't even tell me he
downloaded HTTPAPI until much later) I'm not sure how I could've taken
that attitude with him.
Indeed, I was speaking on HTTPAPI at RPG World. At the time, Cozzi ran
that conference in a partnership with Jon Paris and Susan Gantner.
(Today that's no longer true, it's purely Cozzi's conference. The group
decided they didn't want to work together anymore.) I developed this
session and gave it at this conference. By the time the next RPG World
came around, Cozzi told me that my session wouldn't be needed anymore,
since he had his own.
The library problem was true. Sort of. I originally had LIBHTTP
hard-coded into the system. If you wanted to use a different library,
you'd have to do a PDM option 25 and search out the 'LIBHTTP' string and
replace it with whatever you wanted to use. (Or do the same thing with
the FNDSTRPDM command.)
I'm not sure how Cozzi came to the conclusion that searching for a
string and replacing it was more work than writing his own tool from the
ground up. And I'm not sure why, if he felt so strongly about it, he
couldn't do the FNDSTRPDM, replace it with a variable instead of a
hard-coded reference, and contribute that back to the project. Surely
that would've been easier than writing his own tool from the ground up?
But, he didn't even mention it to me. Didn't even provide me with
feedback telling me what he disliked about the project. Made no effort
at all to improve it.
Nowadays, of course, you can specify any library that you like, or
*LIBL, when you install HTTPAPI. The install program for HTTPAPI asks
you what you want to do, and gives you as much freedom as it can. That
happened because someone else (not Cozzi) mentioned to me that it'd be
nice to have. It took me about an an hour to change it and release it
to the public.
I can only conjecture that there was another reason why he wanted to do
this. If he contributed to HTTPAPI, and improved it, it'd still be my
tool, right? It'd still have my name on it, and I'd still be credited
for it. My impression of Bob is that having credit for the work would
be very important to him.
You'll note that iSockets isn't free, either. Not free as in freedom,
at any rate. He'll give you the compiled objects for free, but if you
want the source, it costs money, and it remains his property. HTTPAPI is
free as in freedom, as well as free of charge.
The other thing that's always bugged me about iSockets is the confused
terminology. iSockets is a tool to simplify creating TCP connections.
It doesn't support all of the things that sockets can do, only what TCP
can do. It's touted all over as an HTTP tool, but it doesn't do HTTP --
you have to do that yourself. In his discussion of "what iSockets can
do" on the iSockets web site, it points you to a web page dedicated to
the HTTP protocol, even though his tool doesn't implement HTTP, it
leaves it up to you to implement it (which is perhaps why you need to go
to that site) But, he never even mentions that you could use iSockets
to write other apps like e-mail, telnet, FTP, or the hundreds of other
tools that run over TCP. I've always wondered if he thinks sockets are
purely for HTTP, and doesn't realize that they're used for all internet
applications? He has a similar confusion about a URL vs. a host name.
His routine called OpenURL doesn't actually open an URL! It establishes
a TCP connection to a host name. To implement the other parts of the
URL (the protocol, directory name and file name, for starters) you have
to write your own code.
He also refers to it as a "free web services library". It's not a web
services library. It doesn't understand SOAP or WSDL or anything like
that. It doesn't even understand HTTP -- you have to code the HTTP
yourself. Could you use iSockets to write a web services client? Sure.
But it'd be YOU writing the web services tool, and using iSockets as a
piece of it. It isn't, by itself, a web services tool.
But, anyway... what am I quibbling for? If he uses his name (he's a
bigger name than I am) and the fact that his software is sort-of free,
and other stories to compete with me, and take away all of my
"customers", what does it hurt me? I don't make money on this.
Jones, Simon wrote:
> "Heck, look at iSockets (www.isockets.net) it is a SOCKETS wrapper
> *SRVPGM that I wrote a couple years ago. Why did I write it? Because
> the one that "everybody" suggested I use, written by a noted Magazine
> Author, was so ridge that I couldn't install it into anything other
> than the library the author initially used. Which meant I had to add
> yet another library to my library list to, compile it, to run it and
> to allow the end-users to use it. The attitude of "you've got the
> source, and it was free so change it yourself" is sort of strange and
> only dampens the desire for others to use it. And this one was
> written by a relatively good developer."
> At least he thinks Scott is a 'relatively' good developer, although
> quite what you need to do to be an 'actually' good developer is beyond
> me at the moment!
This is the FTPAPI mailing list. To unsubscribe, please go to: