On 13-04-20 04:20 PM, Lex Trotman wrote:
My experience based on my consulting customers is that they run old
versions or Red Hat mostly, with only security updates added.  Even point
releases are viewed with suspicion, that is likely to mean they have to
hire me to re-compile their crusty old business apps to new libraries since
their own devs don't understand them. :)


On 21 April 2013 05:13, Matthew Brush <mbr...@codebrainz.ca> wrote:

On 13-04-20 09:52 AM, Nick Treleaven wrote:

On 20/04/2013 16:06, Matthew Brush wrote:

On 13-04-20 05:59 AM, Harold Aling wrote:

On Sat, Apr 20, 2013 at 1:40 AM, Lex Trotman <ele...@gmail.com> wrote:

Agree with Thomas, we should go to 2.24 for windows since we supply the
bundle, Linux doesn't need to go so far, for eg what GTK did the
current red
hat and suse enterprise versions release with, expect 2.18 or 2.20.


I might be stepping completely out of line with this remark: Do people
with dusty old GTK's use bleeding edge Geany?


Yes



It is rumored that some percentage (remember 0 is a percent :) of users
are apparently stuck on ancient enterprise distros at work, need to use
the latest bleeding edge Geany at all times, and are able install it but
not GTK+ from source into $HOME. Or so the story goes :)


Nobody knows how, let alone is allowed, to modify such fundamental
software, even in their own home dir.


IIUC on Unix-ey systems, your home directory is completely writable by you and using the environment you can compile, install and run any software you want. I'm no sys-admin though.



I imagine building newer GTK/GLib from source with only old dependencies
would be a pain in the arse, having to grab and build all the newer
dependencies too. It's also a pain for most users to install and manage
multiple versions of libraries. Geany is actually very easy to build as
it only requires Gtk/GLib.


It is a pain, but I'm more skeptical about what percentage of users are in
this situation. Just thinking about the specific criteria you'd have to
meet to be in this minority of users:

* Running old but still supported enterprise distro


tick

Tick what? Let me get this straight... you are running an old but still supported enterprise distro...


* No ability/access to install a separate system GTK+


tick, tick


And you don't have root/admin rights...


* Aren't satisfied with the Geany in the repos


0.18 or older, no way, tick

And you can't run any release of Geany from 0.00 to present that uses GTK+ < some proposed newer version...



* Need to run the absolute latest Geany from release or Git


Release yes, tick


And arbitrarily need to run bleeding edge Geany...


* Aren't willing to compile/install GTK+ stack into $HOME but are for
Geany, but not for most other popular GTK+ software which requires newer
GTK+.


You already said that, tick again


And aren't able to compile a series of dependencies that takes maybe an afternoon for the first time...



[...]

* Not using lots of other GTK+ software that requires newer GTK+


Whats a GTK?


That library that the GIMP and Pidgin and many other more popular programs use which you'd have to install a newer GTK+ library to use.

For windows there is less of a problem since we, and every other app,
supply a bundle including GTK.  So the windows release can go to 2.24 (the
new recommended release) without too much pain.



So for some possibly insignificant number of Geany users (possibly even
approaching 0 users), who still technically have a path to use latest
Geany, albeit with some hassle (building GTK stack), we
(contributors/developers) have to:


The path of building GTK does not exist in corporate, it is not allowed,
and web app designers don't know how to do it.


If you can't build GTK+ in $HOME it's unlikely you can build Geany there.

And why, if you're a web-app designer do you need to install the absolute latest Geany release, heck, you're probably running OSX if you are one, and IIRC macports and friends stay pretty current on GTK+ versions.




* Quadruple check every GTK+ function we call, referring to out of date
documentation in some cases, to ensure we don't use API from the last 4
years or


tick


Tick what? It's a PITA, some old API functions don't even have a "since" version.


* Ensure all new code using API from the last 4 years is #if'd-out so that
they work with both latest and ancient GTK+.


tick


Tick what? It's a huge pain to write stuff twice, once for "ancient" GTK+ and once for regular GTK+.


* Maintain more stuff in the gtkcompat.h header to fake out a bunch of
core and plugin code so we can also build against latest GTK+.


tick


Tick what? It's already caused troubles and is only a band-aid for the inevitable.


* Have a separate old GTK+ on our dev. machines besides modern ones to
test all changes against (although I don't do this personally).


no, but the nightly does for us,


Which makes Enrico or poor user notify us days later of what we broke.



IMO, the benefits to current and potential contributors and users of
updating to a relatively modern min. GTK+ 2 version outweighs the negatives
to some unknown small percentage of users.


Well, who knows what percentage?  We are talking real users, they will not
be on this list, and might not be on the user list.  They just use
software, they are not involved with it like we are.


Exactly, and I'd wager the *vast* majority of Geany's users are:

1) Running Windows, grab latest setup.exe from website
2) Running Linux, grab latest package from repos
3) Running OSX, grab latest package from macports (or friends)
4) Compile from source using a modern distro.

Even if 1 in 1000 of Geany's users fit your "tick" list, that's 0.1% of all users, and I think even that might be a stretch (based on anecdotal evidence watching the bug tracker, mailing list and IRC).



Even Puppy Linux has had GTK+ 2.24 for the last two releases[1].


And that is irrelevant, a more effective question is, what does Red Hat 6.0
have?  (thats a real question, my quick look at their website couldn't
identify what it is)


It's not irrelevant because Geany is the default text editor on Puppy linux (last I checked) and it used to be one of the reasons not to bump GTK+ versions.

Red Hat 6.0, released in 2010, is still currently in its phase one support
and is still considered bleeding edge by many corporates, Red Hat 5.0 is
still supported but is outside its phase one so I think its safe to ignore
it now.


Wasn't it you who mentioned the trouble of supporting more users in another thread? Why should we make the same compatibility commitment as a publicly traded multi-million dollar corporation? I can assure you if some mega-corp comes along and offers me 20 grand to backport some later fixes, I'll be happy to do so.

I definitively agree that there's someone/a few people out there, in this strange self-imposed situation that are mentioned above, but I'd bet good money it's a such a small minority of Geany's users, possibly even less than 1%, to be worth bending over backwards for.

Cheers,
Matthew Brush

_______________________________________________
Devel mailing list
Devel@lists.geany.org
https://lists.geany.org/cgi-bin/mailman/listinfo/devel

Reply via email to