From kde-core-devel Fri Dec 30 02:26:15 2005 From: Gregory Hayes Date: Fri, 30 Dec 2005 02:26:15 +0000 To: kde-core-devel Subject: Re: platform independent kdeinit/klauncher (was KDE4's IPC) Message-Id: <9f573ec50512291826o5aa943aev6c5a4166b43b0bda () mail ! gmail ! com> X-MARC-Message: https://marc.info/?l=kde-core-devel&m=113590958529236 MIME-Version: 1 Content-Type: multipart/mixed; boundary="------=_Part_2966_8461320.1135909575852" ------=_Part_2966_8461320.1135909575852 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline I had to follow up to this, because it is an excellent example of why Linux desktop developers shouldn't ever concern themselves with the so-called "corporate desktop." I have spent my entire career working in large corporate environments, for both consumer and IT industries. In that time, I have never seen anyone intentionally transition (for the sake of argument) 100,000 users from something they understood to something new without extremely compelling reasons. Without going into much discussion of ROIT, the basic case is that a transiting of your entire "interface" for doing business (i.e. moving to a Linux desktop) requires a _lot_ of additional work. This includes retrainin= g of all of the affected users and their support staff (bye-bye MSCEs ;), re-documenting, re-designing work flow, adding interoperability with archai= c software, intense security requirements, ports of custom software, replacements for special Windows software, and the ability to "blame some external vendor" to list a few. You will also have a number of people who can't, or won't, transition at all for various reasons. These people usuall= y play golf with the CFO, and they will get special exemptions from moving over. This will result in trying to support a dual environment, until the "rest of the world" catches up and makes everything interoperate with x-desktop technology on Linux. If you can get past all that, then you face the real problem. Corporations, when dealing with internal end users, are not interested in "good software" (by the standards of most people on this list), they are interested in "simple tools". While there are cases where both can be one and the same, there are many more where it means "bare bones" or crippled. As some of the Linux flagship corporations get past the server penetration point, you will begin to see people trying to make some kind of "corporate desktop". When you do, you will recognise it, because someone pulling the purse strings will have asked them to reduce the functionality to the bare minimum to do the job and/or placed some crazy requirement of compatibility with x-technology. While it can be bantered that it has been done by x enterprise, or so-and-so seriously considered it -- I have another idea, lets design a desktop *for Linux*. Lets use the best tools we available on the platform, and target the people who are actually using Linux today (i.e. not these "someday maybe" corporate folks). Once we get that knocked out of the ballpark, then we can worry about other things. Greg - On 12/29/05, Stefan Teleman wrote: > > On Thursday 29 December 2005 05:41, Andras Mantia wrote: > > > Not really. Can't be used by anyone outside Qt/KDE. > > > > I don't get it. Is it meant to be used outside at all? (Yes, somehow vi= a > > Rudy in the future). > > i am the CTO of Big Corporation, Inc. i want to standardize my company's > desktop to KDE. 10000 desktops in all. i have some very complex internall= y > developed applications which are the money-makers in my company. KDE is > very > nice, i love it, that's why i choose it, but it does not make me any mone= y > for my business, my business makes money selling green widgets. > > i have to rewrite our internal applications and integrate them into KDE, > which > i am perfectly wiling to do, since KDE will be our brand new shiny > standard > desktop. > > my company's middleware infrastructure is standardized on CORBA. our > internal > applications must be integrated into KDE, therefore i must have a way of > bridging KDE and CORBA. i am the customer, and this is a make or break > feature requirement, because our internal applications receive real time > async notifications from our worldwide sites, and these notifications may > take some further action at the receiver's end, depending on what they > are. > we currently achieve this with CORBA callbacks. > > how do i integrate KDE into our standardized CORBA environment ? if the > answer > is that i have to spend money developing CORBA bindings to DCOP or DBUS, > then > i'm sorry but i will forego KDE and look at other alternatives which have > CORBA support, or at least have a way of easily integrating CORBA > bindings. > my company does not make money developing open source software. we make > money > selling green widgets. > > i was personally involved in a discussion like this, about KDE, 2 years > ago. > the only thing i have changed is the green widgets part (this company doe= s > not make money selling green widgets, but something else). the outcome wa= s > that this company looked somewhere else. > > good luck to me trying to bring up KDE again next year. they won't even > pay > attention. > > --Stefan > > -- > Stefan Teleman 'Nobody Expects the Spanish Inquisition' > steleman@nyc.rr.com -Monty Python > ------=_Part_2966_8461320.1135909575852 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline     I had to follow up to this, because it is an excellent e= xample of why Linux desktop developers shouldn't ever concern themselves wi= th the so-called "corporate desktop." I have spent my entire care= er working in large corporate environments, for both consumer and IT indust= ries. In that time, I have never seen anyone intentionally transition (for = the sake of argument) 100,000 users from something they understood to somet= hing new without extremely compelling reasons.=20

    Without going into much discussion of ROIT, the = basic case is that a transiting of your entire "interface" for do= ing business (i.e. moving to a Linux desktop) requires a _lot_ of additiona= l work. This includes retraining of all of the affected users and their sup= port staff (bye-bye MSCEs ;), re-documenting, re-designing work flow, addin= g interoperability with archaic software, intense security requirements, po= rts of custom software, replacements for special Windows software, and the = ability to "blame some external vendor" to list a few. You will a= lso have a number of people who can't, or won't, transition at all for vari= ous reasons. These people usually play golf with the CFO, and they will get= special exemptions from moving over. This will result in trying to support= a dual environment, until the "rest of the world" catches up and= makes everything interoperate with x-desktop technology on Linux. If you c= an get past all that, then you face the real problem. Corporations, when de= aling with internal end users, are not interested in "good software&qu= ot; (by the standards of most people on this list), they are interested in = "simple tools". While there are cases where both can be one and t= he same, there are many more where it means "bare bones" or cripp= led.=20

    As some of the Linux flagship corporations get p= ast the server penetration point, you will begin to see people trying to ma= ke some kind of "corporate desktop". When you do, you will recogn= ise it, because someone pulling the purse strings will have asked them to r= educe the functionality to the bare minimum to do the job and/or placed som= e crazy requirement of compatibility with x-technology.=20

    While it can be bantered that it has been done b= y x enterprise, or so-and-so seriously considered it -- I have another idea= , lets design a desktop *for Linux*. Lets use the best tools we available o= n the platform, and target the people who are actually using Linux today ( i.e. not these "someday maybe" corporate folks). Once we get that= knocked out of the ballpark, then we can worry about other things.

= Greg
-


On 12/29/05, Stefan Teleman <steleman@nyc.= rr.com> wrote:
On Thursday 29 December 2005 05:41, Andras Mantia wrote:
> > Not r= eally. Can't be used by anyone outside Qt/KDE.
>
> I don't get = it. Is it meant to be used outside at all? (Yes, somehow via
> Rudy i= n the future).

i am the CTO of Big Corporation, Inc. i want to standardize my comp= any's
desktop to KDE. 10000 desktops in all. i have some very complex in= ternally
developed applications which are the money-makers in my company= . KDE is very
nice, i love it, that's why i choose it, but it does not make me any mo= ney
for my business, my business makes money selling green widgets.
<= br>i have to rewrite our internal applications and integrate them into KDE,= which
i am perfectly wiling to do, since KDE will be our brand new shiny stan= dard
desktop.

my company's middleware infrastructure is standardi= zed on CORBA. our internal
applications must be integrated into KDE, the= refore i must have a way of
bridging KDE and CORBA. i am the customer, and this is a make or break<= br>feature requirement, because our internal applications receive real time=
async notifications from our worldwide sites, and these notifications m= ay
take some further action at the receiver's end, depending on what they = are.
we currently achieve this with CORBA callbacks.

how do i int= egrate KDE into our standardized CORBA environment ? if the answer
is th= at i have to spend money developing CORBA bindings to DCOP or DBUS, then
i'm sorry but i will forego KDE and look at other alternatives which ha= ve
CORBA support, or at least have a way of easily integrating CORBA bin= dings.
my company does not make money developing open source software. w= e make money
selling green widgets.

i was personally involved in a discussion= like this, about KDE, 2 years ago.
the only thing i have changed is the= green widgets part (this company does
not make money selling green widg= ets, but something else). the outcome was
that this company looked somewhere else.

good luck to me trying = to bring up KDE again next year. they won't even pay
attention.

-= -Stefan

--
Stefan Teleman      &nbs= p;   'Nobody Expects the Spanish Inquisition'
steleman@nyc.rr.com &nb= sp;            =             -Mo= nty Python

------=_Part_2966_8461320.1135909575852--