For those who like to write code why would you stick on with QCT (not r&d) - when all there is to do is port from one chip to another, triage the same issues on every chip, and next to no chance of writing new code?
15 replies (most recent on top)
OP here. I disagree with the LOC metric too, the most concise code is the most elegant and easiest to maintain.
But what features and design are you talking about?? Most of the 'design' is - how to architect the code such that tweaking it for the next chipset needs the least effort.
I disagree with LOC metric. LOC doesn't mean shit. People spew out shit code all day long. And most of us are debugging the shit others spew out all day long.
I don't care about LOC because I'm an Engineer, not a programmer. I design and implement features. I make it work. Coding is one part of what I do to implement the feature. It is not the LOC that matters. It is a proper design, an engineering approach, and the whole process to get this feature mature and working well. Analysing, debugging, improving the systems design. LOC=cheap. Making it work=priceless!!
I have worked at both places. Corp R&D has only 1 or 2 real projects that will see the light of the day, the remaining are very obviously a waste of QC money (like neo or robotics project) and just a boondoggle to keep the engineers there going, but nobody will openly admit. Moreover CR&D is run by a 'good old boy network', and it is very difficult to break in especially at the senior staff and above level, if you haven't been there a very long time, or run afoul of the above mentioned good old boys. In QCT it is much easier, and folks are much more open minded, IMHO..
IN QCT, while the coding is generally less the CR&D, the code that I wrote in QCT got ported to multiple generation of modem, and still get a lot of questions on it, from other developers. So in a sense it is pretty rewarding. There also lots of new features, that are built on top of existing very mature code, so it is a challenge to bring it up with minimal disruption, and requires very detailed knowledge of the existing massive code base. Bringing in a new feature requires lots of planning, as well coordination with multiple parties.
IMHO, I think the soft skills that you learn in QCT, will prepare you for more real word, while CRD coding in general looks better on a resume.
About 500 LOC/week, tested and functional, day in and day out used to be the norm.
I heard that NASA produces about 1 LOC/day/developer, but then again, human lives and very, very expensive machinery depend on that code.
Let's see, in the past 6 months, oh darn, I certainly wrote less than 500 LOC, because I was too busy fixing all the other developers bugs from the past. I guess I'm no SW engineer anymore. I'm just a dumb looser SW maintenance dude who knows the system in and out better than the developers who spew out this crap day in and day out. And the worst of it all is, I like it!
I have to agree with the OP but my experience hasn't been much better at other places. It really seems hard to find a place that has something new and exciting. Too many legacy code bases out there and your job is to shoehorn some new feature into a fragile pile of crap code.
I have to agree with the OP but my experience hasn't been much better at other places. It really seems hard to find a place that has something new and exciting. Too many legacy code bases out there and your job is to shoehorn some new feature into a fragile pile of crap code.
I have to agree with the OP but my experience hasn't been much better at other places. It really seems hard to find a place that has something new and exciting. Too many legacy code bases out there and your job is to shoehorn some new feature into a fragile pile of crap code.
Q is a hardware company
I don't know. But the guys I've been interviewing from QCT for software positions sure don't know how to write software worth shit...even the ones that claim they know Object Oriented Design.
You should do a quick check of the number of lines of code that you actually write per year (use something like git blame, which is lines committed, which will be > than lines you wrote, but still a starting point). If you are doing < 5-10K LOC per year, you are f'ed. You should be looking to do at least 10-20k+ LOC created (not touched) per year. Anything less, and you aren't really a SW dev. You need to do actual code creation to grow your skills otherwise you are just maintainer.
QCT( and Q in general) is a trap for s/w engineers...Very minimum learning and too much bureaucracy..
QCT does not need that much sw engineering.They mostly( >80%) need porting/integration and test folks to support third party OS to support customers sell their chips.<~20% s/w ( who know embedded s/w) folks do real sw work to get chip up and running ( and optimizing some aspects of it)....so it depends which group you are working.Also remember QCOM is not a S/W company ( they tried in past but did not work out),SW is supporting role to get our hw (chips) out of the door and bring in money....Apple,Goog,Facebook etc are different beast.
New features! New registers! New settings!
Care to elaborate?
Btw I was there for 5 years and heard the same sentiment from peers who also left. So are/were we all naïve?
What a naive question/statement/comment.