Mail Archives: geda-user/2016/02/28/23:39:58
On Sun, 28 Feb 2016, John Doty wrote:
>>>
>>> Except that if you watch this list, the most common topic is problems with pcb. A significant subset is problems with pcb that people expect geda-gaf to work
>>
>> That's because what pcb is trying to do is harder.
>
> Is it that the problem is intrinsically harder, or that the approach of implementing a function for every use case ad hoc makes the problem harder?
>
I know this is a rehash, but a "short" summary in case of someone new is
reading the thread...
Unlike John, I never sent anything to space. But unlike John, I am an
active user of both pcb (fork) and gschem. My flow is not exclusively
gschem->pcb. I also hacked the code of both pcb (a lot) and gschem (a
little) and the accompanying tool chain.
My experience is exactly the opposite. TLDR: geda/gaf/gschem is more
broken than PCB.
Long version:
PCB: hackish and limited, most of the limitations stem from design, but it
actually does its job surprisingly well on the common daily stuff. When I
need something new and strange, it's relatively easy to add as long as it
doesn't try to totally change the concept of layers.
Most importantly: the community (devs and users) are well aware and open
about the shortcomings in the design and are considering/working on to
change them. PCB tries to evolve using multiple different techniques. It
tries to find out when an old design decision is bad and it admits it and
it tries to leave it behind. PCB is brave enough to discuss and touch
even core concepts of its design. If something is not fixed or not fixed
soon enough is usually because of lack of developer time. My impression is
whether pcb will survive another decade mostly depends on whether enough
manpower is put into finding and fixing the right problems.
gschem: hackish and limited, most of the limitations stem from design, but
it actually does its job surprisingly well on the common daily stuff. When
I need something new and strange, it's a PITA to add, because some of the
most trivial and basic concepts are missing by design. Despite of the
rugged effort of some hardcore anti-pcb users, none of these are related
to the pcb flow or non-toolkit approaches, rather things like "objects can
not be identified", we lie "networks (mostly) don't exist", and that
"scheme is so great that everyone must learn it", etc. The common in them
is the fact that gschem tries to be highly generic at some parts while it
is simply not generic enough (or at all) on other parts. Note: I am
talking about generic design/concept and inrastructural stuff, not
about a bag of random UI features.
Most importantly: the community (devs and users) are either aware of this
and then labeled "anti-toolkit", "anti-gschem", "pcb users" or they are
the Few True Geda Users, for whom geda is perfect and who are totally
blind to any design flaws. Thus there's no much chance to even discuss
design flaws, yet alone fix them. My impression is whether geda/gschem
will survive another decade depends on whether pcb keeps it as an almost
exclusive input. Not because pcb is so great and flawless - it's more like
"how far gschem, as a standalone project, would continue from now if we
didn't have pcb". It's because gschem is carrying some old and wrong
design decisions that blocks a lot of potential roads to progress and
there seems to be no chance removing them.
No offense meant. Developers are investing a great deal of work in both
project. Both projects are mature and useful. It's just that while pcb is
getting bashed daily, it's gschem that seems to be much more stuck with
hard to fix (or even discuss) design decisions from the past. I do not
want to discourage developers even if I don't agree with the direction
which some of them take.
NOTE: I don't want to start a flamewar (it's already raging on). I don't
want to contribute to the flamewar (too much), so I probably won't answer
to replies except something radiacally new appears. My points in details
are known (look in the mailing list archives or in my devlog).
My intention with this mal is...
Many of us try to restrain themselves and avoid defending pcb from
unreasonable and often baseless attacks or avoid trying to discuss
geda/gschem design flaws as they turn into senseless flame wars within 2
mails. In the same time some users keep coming back claiming pcb is all
bad and gschem is perfect. The sole purpose of my post is to state
that's not the ultimate truth and there's another opinion on this.
Regards,
Igor2
- Raw text -