embeddedgurus.com: "An open letter to the developers of the MPLAB IDE"

[nctnico]

I have been using it exclusively for all my software development for about 10 years now.

For such an "expert" on Eclipse, you seem to have more than your fair share of troubles getting CoIDE (=Eclipse) to work.

I think you are mistaken me for someone else. I never tried, used or mentioned trying / using CoIDE.

[westfw]

If you look at the development history of GCC, the overwhelmingly largest amount of work has been commercial. It's also incorrect to say that Microchip have done "nothing".

I find it hard to believe Microchip added features to GCC which improved MIPS support in general. Even though MIPS is not a general purpose core

And the pre-microchip MIPS support in gcc was contributed by ??  Mostly commercial entities, wasn't it (SGI, MIPS, and assorted licensees?)  I guess maybe it gets "fuzzy" when architecture gurus found companies while retaining professorships, and get their grad students to implement stuff.  (not that this is at all "new", or related to the OSSW "movement."   BASIC came from Dartmouth, etc)

And MIPS is a fine "general purpose core."  Just not a very successful one, recently.  For a while, there were LOTS of MIPs CPUs in workstations, handhelds, gaming systems, routers, and so on.  ARM and Intel just sorta rolled over them.

[SirNick]

does anyone really serious, motivated or commited spending thousands of hours for a "quality" FOSS?

Yes.  While anyone can come up with a laundry list of crappy (free or commercial) programs, let's not forget that many devices (PC and non-PC) use Linux or BSD as the kernel, run a BSD network stack, use an HTML rendering engine based on KHTML (which became WebKit, which became Chrome and Safari), use OpenSSL for transport layer encryption, and were compiled with GCC.

There are significant efforts put behind FOSS, and some of the most prolific software in use at this time is, in fact, FOSS.  Or uses it to do considerable work behind the scenes.  The smart hardware vendors take advantage of that fact, publish their specifications, and let the eager code monkeys take on the lion's share of development for free.  Coders get a fun project and working hardware, vendors get free development, users get free tools.  I really can't understand how this model leaves anything to complain about?

...let alone making a effort to make it work at the first install in any type of OS targetted at.

One of the most user-unfriendly installs I've ever had to work my way through was for Oracle's database product.  Let's just say, it's not quite a free product.  MySQL (also now under Oracle's care, but a FOSS product) on the other hand is much easier to install and configure.  Now these two products aren't exactly equivalent, but hey if we're throwing around generalisms with regard to the polished-ness of software...

blah blah KiCAD yada yada

KiCAD is crap.  On behalf of the OSS community, "We're sorry about that."  Who knows, it may turn into a brilliant project some day.  At least for now, forget KiCAD.  There's just a lot of buzz right now because the market is aching for a decent tool that can be made ubiquitous without any financial consideration.  It's a void that is collecting matter, and KiCAD was the first piece of detritus to show up.  It will get better.  (The situation.  Not KiCAD necessarily, as it is crap.  Currently.)

[Mechatrommer]

users get free tools.

but users dont get free iphone or samsung galaxy. that was my point as i saw someone moan about paying 1K for mplab whatever just to get the optimized gcc unleashed, you are paying for mplab not gcc. Al-tee-um is crazy expensive like a car, even 5K yearly just to get bug fix support, understood, its closed source delphi. mplab is part close and part open source is 1K?, yeah what we are compaining about? carry on, economy will balance itself.

[westfw]

I really can't understand how this model leaves anything to complain about?

"Invasive species" - killing off a bunch of smallish sw vendors who sold sw.  I start to worry a bit about platforms that have only gcc as assembler/compiler.  They're some starry-eyed dreamer's whim away from having an unsupported mess on their hands.

[nctnico]

Open Source software is never unsupported. You have all the sources available so you can provide your own support or (more common) pay someone to make the changes you want.

[mrflibble]

I start to worry a bit about platforms that have only gcc as assembler/compiler.  They're some starry-eyed dreamer's whim away from having an unsupported mess on their hands.

Mmmh, that's a valid concern yes. But I think that if it really came to that unsupported mess (because said dreamer did something whimsical) then you would no doubt have a fork in no time. Same as on the commercial side of things where "Oracle buys X", followed by "Oracle fucks up X", followed by "X development is forked".

[westfw]

Open Source software is never unsupported. You have all the sources available

Having source available so that you can "fix it yourself" or "hire someone to fix it" is certainly much better than NOT having the source available, but I still claim that it is "unsupported."

Consider, say, "WinAVR"; the package including avr-gcc and associated tools for windows cli use.
The last "release" was in 2010, based on avr-gcc 4.5 (I think.)  If I were somehow dependent on that version, I would be very nervous.  (Sound far-fetched?  Arduino was in exactly this position - they were based on 4.3.x Crosspack/WinAVR, and very version of avr-gcc between that and 4.7.2 or so had some signficantly-disabling bug WRT Arduino.  They're still agonizing over 4.8.x (there's no winavr package, but the Atmel command line tools are almost equivalent.))


If windows 8.2 stops supporting 2010-era MinGW command line utilities (which seems "possible"), then I could be quite screwed in a couple of years.

(Not that the situation with commercial compiler vendors is any better, should they make choices that conflict with my desires.  But having only ONE compiler does make it worse.)

[andersm]

(there's no winavr package, but the Atmel command line tools are almost equivalent.)

Atmel hired the WinAVR maintainer to make the Atmel command-line tools.