Experimental Physics and
Industrial Control System

"keith.thorne" <[email protected]> · Thu, 1 Mar 2018 15:00:21 -0600

Dear Matt and company

I will add that there is a modest effort in LIGO working on moving from MEDM to caQtDM. Much of the work is fixing existing screens (before or after conversion) to have proper box sizes for the text blocks for font sizing. We are proceeding slowly, doing it in parallel with existing MEDM. Some operators are getting experience with the Qt editor.

An initial use (of benefit to users) would be deploying caQtDM on tablets (iOS or Android) using the page-download feature and native Qt installation.  This will be much better than the remote desktop session we use now.

- On a related note:  What are the best options to replace Striptool? (also antique X-11, also EOL). 

Sincerely
	Keith Thorne

> On Mar 1, 2018, at 2:34 PM, Matt Newville <[email protected]> wrote:
> 
> James,
> 
> On Thu, Mar 1, 2018 at 10:01 AM, Jameson Graef Rollins <[email protected]> wrote:
> On Wed, Feb 28 2018, Pete Jemian <[email protected]> wrote:
> > At this time, MEDM is an orphan piece of software.  There are no
> > developers assigned to maintain it.  Fantastic that it still works and
> > is still useful.  Great job to its developers!  MEDM's true end of life
> > will coincide with that of MOTIF, which library it uses.  Recently, some
> > have made code changes in response to specific problems, but MEDM has no
> > maintainer.
> 
> Despite it's annoying limitations, MEDM has a lot going for it.  For
> instance, the simplicity of the file format makes it easy to dynamically
> generate screens, which is a very nice feature.
> 
> As Pete and Andrew have pointed out, MEDM also has a lot NOT going for it.   These anti-features include "not supported" and "uses antiquated X library that is itself poorly supported".  We see real problems with MEDM and modern X servers. More obviously, MEDM requires X11, which has long been a significant and painful requirement for folks using Windows and/or MacOS, and is a non-starter for people wanting to use Android or iOS.
> 
> No one else has mentioned this, but MEDM has always been simply horrible in handling fonts.  It uses custom X11 font names that require modifying the X11 font table.  I am not aware of any other software in regular use that makes such a poor choice. A display manager like MEDM is fundamentally a graphical display program. Most of the objects it displays being text. MEDM gets one of the most basic components of displaying text (ie, fonts) absolutely wrong.  It is not the cases that the choice made ~25 years ago was OK, and is now out-of-date: custom fonts for X11 were a horrible choice 25 years ago too.  MEDM has always done fonts wrong and anyone unfortunate enough to be using MEDM is stuck with this problem and very poor fonts to this day.
> 
> MEDM is OK for dynamically generating screens. It includes a quirky but impressive home-built GUI generator. Qt Designer is far better.  It may have a slight learning curve for someone familiar only with MEDM, but it very good and has many more features and is well-supported and widely used in other contexts.
> 
> In short, there is nothing about MEDM that is not greatly improved by caQtDM.
> 
> 
> At LIGO we're pretty stuck with MEDM just because of the sheer amount of
> work required to translate and test all of our existing screens.  This
> problem exists no matter which new system we eventually decide to move
> to, so a requirement for any new system is a good translator.
> 
> The adl2ui translator is very good.  We have translated more than 1500 adl files to ui file and really use these (for the past couple years). The only recurring translation problems have been
>    a) groupings in MEDM do not translate well, and can leave unresponsive areas of the screen.  Ungrouping the adl files fixes this.
>    b) the space used by text boxes in MEDM is not enforced, probably related to its font handling being broken by design.
> 
> There is nothing to recommend MEDM.
> 
> --Matt Newville 

------
Keith Thorne <[email protected]>
CDS Software Engineer
LIGO Livingston Laboratory
Livingston, LA 70754
Phone: (225)686-3168 Fax: (225)686-7189

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System