Software that makes placemats

[DRT]

Also beware that PermittedPackingStyles is being changed, a new version being likely to appear soon.

The anticipation is almost unbearable.

[Glenn E.]

I’ve slightly shrunk the default margins, from 36 pt = to 30 pt, which might have that effect. Let me test.

Edit: no. Could it be 8 glasses rather than 10, and /RectangularAlternateNudgeIf2pt (a recently constructed base style) rather than /PseudoHexagonal?

I could be mis-remembering 8 glasses in 3-2-3 arrangement. It would be from long enough ago that I doubt /RectangularAlternateNudgeIf2pt is the answer. I'll go back through my saved placemats when I get home tonight, but thought I'd ask since I'm attempting to put together another set of placemats at work today.

Records show that it was indeed 8 glasses in 3-2-3 arrangement that I was mis-remembering as 3-4-3. I have produced that twice on USL and once on USLegal. I have also produced 3-3-3 on USLegal, but when moving to 10 glasses the packing spreads to provide better room and switches to 3-2-3-2 (top to bottom).

[jdaw1]

Conclusion: happiness?

[Glenn E.]

Conclusion: happiness?

A more accurate description would be resignation.

I'm not fond of the layout with 10 glasses on /USLegal, but can accept it if it is the best that we can do. I realize that 10 glasses is approaching the physical limit for /USLegal, but I prefer more obvious rows so that it is easier to see the flights. If it were possible to force a 3-4-3 layout (which assumes that doing so fits on the paper) instead of the current 3-2-3-2 (top to bottom), that is what I would probably always choose for 10 glasses on /USLegal.

As is probably obvious, I don't have a problem with a much more densely packed placemat than what seems to be preferred in London.

[jdaw1]

Then delete from PermittedPackingStyles the arrangements you don't like. You have control.

[jdaw1]

Those deletions done, also def both RowsMinNum and RowsMaxNum to 3.

[Glenn E.]

Then delete from PermittedPackingStyles the arrangements you don't like. You have control.

I do.

/PermittedPackingStyles [ /PseudoHexagonal ] def

Is that not the correct style?

[Glenn E.]

Those deletions done, also def both RowsMinNum and RowsMaxNum to 3.

Aha! Most excellent. Yet again, proof that following this thread is good.

[Glenn E.]

Then delete from PermittedPackingStyles the arrangements you don't like. You have control.

/PermittedPackingStyles [ /PseudoHexagonal ] def

Those deletions done, also def both RowsMinNum and RowsMaxNum to 3.

/RowsMinNum 3 def
/RowsMaxNum 3 def

Confirmed. Much happiness.

[jdaw1]

For the prevention of future confusion, the PermittedPackingStyles parameter is about to be re-arranged, to a wider range of patterns, with control that both allows more detail, and is more logical.

And the incorrect name /PseudoHexagonal will be changed to /Diamonds, as shorter and more accurate.

But, broadly, the overall approach won’t change. The machine will go through the allowed patterns, choosing the one with the largest radius, that is, the most densely packed.

Broadly? The user will be able to specify, for each pattern, an ImprovementPointsMin and ImprovementProportionMin. The thinking is that these can be specified in patterns the user likes less (for me, /RectangularAlternateNudge), which will be used only if materially better than the others. Materially? For this base style, the defaults might be 2 and 0.02, so to be selected the radius must be better by both 2pt ≈ 0.7mm, and by 2%.

The likes of RowsMinNum and RowsMaxNum will be specifiable at the level of a pattern, rather the globally.

[jdaw1]

the incorrect name /PseudoHexagonal will be changed to /Diamonds, as shorter and more accurate.

/Gaia is also a nonsensical name. Please suggest a name for the class of styles of which the following are canonical examples.

[PhilW]

/Gaia is also a nonsensical name. Please suggest a name for the class of styles of which the following are canonical examples.

/Arch

[jdaw1]

/Arch

This has a flat top, whereas an arch is curved. But the idea of an architectural term is excellent.

[PhilW]

/Arch

This has a flat top, whereas an arch is curved. But the idea of an architectural term is excellent.

I thought that arches could theoretically be flat (or variety of other shapes), with curved simply being the most common form? i.e. architecturally, does it not simply require to be an opening where the top is weight-bearing?
Otherwise perhaps /PseudoArch as per previous /PseudoHexagonal.

[jdaw1]

Fair, but insufficient. The under-development pattern currently known as /SemiEllipse has the curved nature that feels more ‘arch’y. So it would be better to capture more decisively the flatness of the top.

[djewesbury]

I merely note the following;

John Ruskin notes many different shapes of arch in his seminal The Stones of Venice.

An Orange arch is flat-topped, albeit with some sort of illustration of a curve on the lintel.

Lintel might in fact be appropriate although that is only the top straight part.

Or PseudoLintel or PseudoDoorway. Or OpenRectangle.

Ah - a proscenium arch is flat-topped. Proscenium?

[jdaw1]

Proscenium?

Pretentious? Qui, moi?

[djewesbury]

Proscenium?

Pretentious? Qui, moi?

Alright then, PseudoProscenium..
I can think of no better peri-architectural term to describe the shape you have presented us with.<sup>1

1</sup> ...with which you have presented us.

[jdaw1]

‟Pretentious? Qui, moi?” was not a criticism. But, for my taste, ‟/Lintel” is currently winning. Maybe it should be /PostsAndLintel, but brevity is desired.

[djewesbury]

‟Pretentious? Qui, moi?” was not a criticism. But, for my taste, ‟/Lintel” is currently winning. Maybe it should be /PostsAndLintel, but brevity is desired.

/PostsAndLintel has a certain ring to it.. And it's immediately clear.

[Glenn E.]

/perimeter

[jdaw1]

/perimeter

Might that be thought to include the bottom edge?

[Glenn E.]

/perimeter

Might that be thought to include the bottom edge?

Hmm. Yes, it probably would.

/DodrantPerimeter?

[djewesbury]

/perimeter

Might that be thought to include the bottom edge?

Hmm. Yes, it probably would.

/DodrantPerimeter?

Et maintenant, qui est prétentieux?

[Glenn E.]

/perimeter

Might that be thought to include the bottom edge?

Hmm. Yes, it probably would.

/DodrantPerimeter?

Et maintenant, qui est prétentieux?

I have no idea what you're talking about.

[Glenn E.]

For the record, I have been following the /PermittedPackingStyles discussion and am happy with the direction it has taken. I've just been busy and so haven't been online in time to make relevant comments.

[jdaw1]

/DodrantPerimeter?

As I don’t understand this, even with Google’s help, no.

[djewesbury]

/DodrantPerimeter?

As I don’t understand this, even with Google’s help, no.

‟They are of divers heights, as of 12, 13, or 14 dodrants, the dodrant being a measure of 9 inches; and some say that an elephant is bigger than three wild oxen or buffaloes.”

[jdaw1]

Failing cogent objection, I propose to name /Arch the base style of which the following are canonical examples.

There is also a need to name the sub-parameter controlling the number of things in the centre. Candidates include /ExtraCircles and /ExtraGlasses.

[djewesbury]

I like these. What about /CentreCircles? Perhaps the Americans won't like that..

[Glenn E.]

/DodrantPerimeter?

As I don’t understand this, even with Google’s help, no.

‟They are of divers heights, as of 12, 13, or 14 dodrants, the dodrant being a measure of 9 inches; and some say that an elephant is bigger than three wild oxen or buffaloes.”

If Wikipedia is to be believed, dodrant is the Latin prefix for 3/4 in the same way that semi is the prefix for 1/2.

[jdaw1]

If Wikipedia is to be believed, dodrant is the Latin prefix for 3/4 in the same way that semi is the prefix for 1/2.

Fantastic. But an ellipse occupies ¼π ≈ 0.785398163397448309615660845819875721049292349843776455243736148076954101571552249657 ≠ ¾ of its bounding rectangle.

So no.

Separately, for every other base style the radius can be found analytically. But not for /Arch, which requires an interval-bisection iteration, and in a loop within that, a more complicated iteration. Given the limits of PostScript’s single-precision arithmetic, I think that the chosen radius is probably correct (that is, maximal) to within 0.01 points ≈ 0.035 mm.

[djewesbury]

If Wikipedia is to be believed, dodrant is the Latin prefix for 3/4 in the same way that semi is the prefix for 1/2.

Fantastic. But an ellipse occupies ¼π ≈ 0.785398163397448309615660845819875721049292349843776455243736148076954101571552249657 ≠ ¾ of its bounding rectangle.

So no.

Separately, for every other base style the radius can be found analytically. But not for /Arch, which requires an interval-bisection iteration, and in a loop within that, a more complicated iteration. Given the limits of PostScript’s single-precision arithmetic, I think that the chosen radius is probably correct (that is, maximal) to within 0.01 points ≈ 0.035 mm.

[PhilW]

Failing cogent objection, I propose to name /Arch the base style of which the following are canonical examples.

There is also a need to name the sub-parameter controlling the number of things in the centre. Candidates include /ExtraCircles and /ExtraGlasses.

Another option for that style: /HalfElipse.

/ExtraCircles not liked, 'Circles' seems the wrong term here; /ExtraGlasses ok or ExtraGlasses<Position> might be better, to allow variants dependent on layout to put an extra group at the defined position, which could either be relative to page e.g. ExtraGlassesCentre or ExtraGlassesTopLeft or ExtraGlassesBottomCentre or relative to design e.g. /ExtraGlassesAtLocus (guessing the former would prove more useful).

[jdaw1]

Another option for that style: /HalfElipse.

The under-development pattern currently known as /SemiEllipse

Preferences? We have /Arch, /HalfEllipse and /SemiEllipse. The centres of the circles do indeed lie on an ellipse, forcing equations to be solved iteratively, but nonetheless the architectural /Arch appeals most to me. But I would defer to others’ strong counter-preferences.

/ExtraCircles not liked, 'Circles' seems the wrong term here; /ExtraGlasses ok or ExtraGlasses<Position> might be better

Currently only two base styles allow extras, and both at the same position. So /ExtraGlasses is currently winning.

[djewesbury]

Preferences? We have /Arch, /HalfEllipse and /SemiEllipse. The centres of the circles do indeed lie on an ellipse, forcing equations to be solved iteratively, but nonetheless the architectural /Arch appeals most to me. But I would defer to others’ strong counter-preferences.

/Arch. Shorter is better. It's also immediately obvious.

Currently only two base styles allow extras, and both at the same position. So /ExtraGlasses is currently winning.

I like.

[PhilW]

Preferences? We have /Arch, /HalfEllipse and /SemiEllipse.

Personally prefer /SemiEllipse but not unhappy with any.

/ExtraCircles not liked, 'Circles' seems the wrong term here; /ExtraGlasses ok or ExtraGlasses<Position> might be better

Currently only two base styles allow extras, and both at the same position. So /ExtraGlasses is currently winning.

ok

[PhilW]

‟Pretentious? Qui, moi?” was not a criticism. But, for my taste, ‟/Lintel” is currently winning. Maybe it should be /PostsAndLintel, but brevity is desired.

Alternative: /Goalposts

[djewesbury]

‟Pretentious? Qui, moi?” was not a criticism. But, for my taste, ‟/Lintel” is currently winning. Maybe it should be /PostsAndLintel, but brevity is desired.

Alternative: /Goalposts

/AssociationFootballGoalposts?

[djewesbury]

You can't deny that /Portal would be quite appropriate..

[jdaw1]

You can't deny that /Portal would be quite appropriate..

It might encourage one Quinta to use the software, at the price of discouraging others.

[Glenn E.]

If Wikipedia is to be believed, dodrant is the Latin prefix for 3/4 in the same way that semi is the prefix for 1/2.

Fantastic. But an ellipse occupies ¼π ≈ 0.785398163397448309615660845819875721049292349843776455243736148076954101571552249657 ≠ ¾ of its bounding rectangle.

So no.

?

the incorrect name /PseudoHexagonal will be changed to /Diamonds, as shorter and more accurate.

/Gaia is also a nonsensical name. Please suggest a name for the class of styles of which the following are canonical examples.

/Arch is my favorite for the other packing style under discussion. But /DodrantPerimeter was meant to replace /Gaia, which was deemed nonsensical. It does, however, fail the length test vs /Gaia.

[jdaw1]

/Arch is my favorite for the other packing style under discussion.

/Arch is decided for the likes of

But /DodrantPerimeter was meant to replace /Gaia, which was deemed nonsensical.

Please explain why /DodrantPerimeter is better than the current favourite, /PostsAndLintel, for the likes of

[Glenn E.]

But /DodrantPerimeter was meant to replace /Gaia, which was deemed nonsensical.

Please explain why /DodrantPerimeter is better than the current favourite, /PostsAndLintel, for the likes of

Neither better nor worse; different. One type of esoterica or another.

Though we probably ought to be attempting to standardize all of the names. If /Arch, then /PostsAndLintel. If /SemiElipse, then /DodrantPerimeter. But in either case, other styles might also need renaming.

[jdaw1]

other styles might also need renaming.

Current state of naming: /RectangularDislocation, /Diamonds, /RectangularAlternateNudge, /Adjusted5, /PostsAndLintel, /Arch, /SquareGrid, /TopRow, /MiddleRow, and /BottomRow.

[Glenn E.]

other styles might also need renaming.

Current state of naming: /RectangularDislocation, /Diamonds, /RectangularAlternateNudge, /Adjusted5, /PostsAndLintel, /Arch, /SquareGrid, /TopRow, /MiddleRow, and /BottomRow.

Of these, only Diamonds, Arch, PostsAndLintel, and SquareGrid are sufficiently descriptive to me. And that assumes that SquareGrid does what I think it sounds like it does.
TopRow, BottomRow, and MiddleRow might also be sufficiently descriptive if they are describing the placement of a single row of glasses on the sheet. The fact that I am not sure indicates, at least to me, that the naming in general isn't sufficiently descriptive.

Diamonds, Arch, and PostsAndLintel are evocative. SquareGrid, TopRow, BottomRow, and MiddleRow are purely descriptive. The remainder are likely attempts to be purely descriptive, but for me are just confusing.

Renaming the first three to OffsetRows, SemiElipse, and Perimeter would make them purely descriptive. (Avoiding the admittedly esoteric use of dodrant- in the process.) I'm insufficiently creative at the moment to come up with evocative names for the other 4 and don't understand the remainder sufficiently to make any suggestions at all.

[jdaw1]

The Chirality of /Diamonds

Some of the /Diamonds arrangements (formerly /PseudoHexagonal) have chirality, that is, they exist in left- and right-handed versions. E.g., this six-glass arrangement is different when reflected.

Which should be the default, and which the /Mirror version?

Other examples:









However, if both the number of rows and the number of columns are odd, there is no chirality:






There is a current rule. When there is chirality, two corners have circles tight into them, and two don’t. My current rule says that the bottom-right corner doesn’t. The thinking was that this might, slightly, reduce the likelihood of a right-hander knocking over that bottom-right glass, which I imagined might be the most vulnerable.

Do any readers have a superior alternative specification of the non-/Mirror default?

[djewesbury]

I always have a slight problem with very tightly-stacked placemats, which is that the first drink of the flight is the furthest away.. Perhaps this is why I'm pleased to see the new /Arch and /PostsAndLintel designs, which give clear access to every glass. Given that /Diamonds exists and will undoubtedly be required at some time, however, I agree that the bottom right corner should be kept clear for right-handers.

[Glenn E.]

The Chirality of /Diamonds

Some of the /Diamonds arrangements (formerly /PseudoHexagonal) have chirality, that is, they exist in left- and right-handed versions. E.g., this six-glass arrangement is different when reflected.

Which should be the default, and which the /Mirror version?

To me, the example on the left is clearly the primary and the example on the right is the mirror.

But the clarity fails when the other examples are perused. I need more time to ponder a solution, though my first impulse is to say that putting a glass as close to the top left corner of the sheet is primary.

[djewesbury]

putting a glass as close to the top left corner of the sheet is primary.

This does seem to make sense at first, but then picture yourself reaching for that first glass at the back of the mat.. It's something delicious, perhaps poured from a Fonseca 1970 mag that you'll be comparing with other formats.. Oh no, your sleeve has just caught the last glass - the Dow 1950! Oh well, it is a pretty colour on the placemat..