Hi folks,
Apologies for the silence: things have been a bit busy.
Test wise we're looking pretty good. We still have a few red lights but I don't see much point in delaying further: some of the test failures are because of VM sensitivities.
Off the top of my head we/I still need to write a changelog, and otherwise a bunch of infrastructural stuff's all that's needed: a new update stream, switching it over, etc. I wrote a ReleaseBuilder script but of course should actually publish it.
I've also nearly finished the new background, and just need to figure out how to clean the edges of the pic to make it tile properly. Given I'm writing this mail on my phone I will have to mail the proposed image separately, at a later stage.
Anyone spot anything missing off the swiki release page?
frank
Frank Shearar wrote:
just need to figure out how to clean the edges of the pic to make it tile properly.
Making Seamless Background Tiles http://graphicssoft.about.com/library/weekly/aa080499p2.htm
On 21 November 2012 14:39, Frank Shearar frank.shearar@gmail.com wrote:
Hi folks,
Apologies for the silence: things have been a bit busy.
Test wise we're looking pretty good. We still have a few red lights but I don't see much point in delaying further: some of the test failures are because of VM sensitivities.
Off the top of my head we/I still need to write a changelog, and otherwise a bunch of infrastructural stuff's all that's needed: a new update stream, switching it over, etc. I wrote a ReleaseBuilder script but of course should actually publish it.
I've also nearly finished the new background, and just need to figure out how to clean the edges of the pic to make it tile properly. Given I'm writing this mail on my phone I will have to mail the proposed image separately, at a later stage.
For better or worse, here it is. Test it with
ActiveWorld color: (InfiniteForm with: (PNGReadWriter readFromFileNamed: 'my/path/to/blur_ulam_1.png'))
Note the strange vertical white-ish line between the tiles. The image is filled all the way to the edge, but the left and right edges differ sufficiently to create the artifact. I'm not sure how to get rid of it, never having attempted to make a tile-able image before.
Anyone spot anything missing off the swiki release page?
That would be http://wiki.squeak.org/squeak/6188.
frank
On 21/11/12 15:53, Frank Shearar wrote:
On 21 November 2012 14:39, Frank Sheararfrank.shearar@gmail.com wrote:
Hi folks,
Apologies for the silence: things have been a bit busy.
Test wise we're looking pretty good. We still have a few red lights but I don't see much point in delaying further: some of the test failures are because of VM sensitivities.
Off the top of my head we/I still need to write a changelog, and otherwise a bunch of infrastructural stuff's all that's needed: a new update stream, switching it over, etc. I wrote a ReleaseBuilder script but of course should actually publish it.
I've also nearly finished the new background, and just need to figure out how to clean the edges of the pic to make it tile properly. Given I'm writing this mail on my phone I will have to mail the proposed image separately, at a later stage.
For better or worse, here it is. Test it with
ActiveWorld color: (InfiniteForm with: (PNGReadWriter
readFromFileNamed: 'my/path/to/blur_ulam_1.png'))
Note the strange vertical white-ish line between the tiles. The image is filled all the way to the edge, but the left and right edges differ sufficiently to create the artifact. I'm not sure how to get rid of it, never having attempted to make a tile-able image before.
Anyone spot anything missing off the swiki release page?
That would be http://wiki.squeak.org/squeak/6188.
frank
Will this do? Done in Gimp in two ticks :-)
On 21 November 2012 16:49, Derek O'Connell doc@doconnel.f9.co.uk wrote:
On 21/11/12 15:53, Frank Shearar wrote:
On 21 November 2012 14:39, Frank Sheararfrank.shearar@gmail.com wrote:
Hi folks,
Apologies for the silence: things have been a bit busy.
Test wise we're looking pretty good. We still have a few red lights but I don't see much point in delaying further: some of the test failures are because of VM sensitivities.
Off the top of my head we/I still need to write a changelog, and otherwise a bunch of infrastructural stuff's all that's needed: a new update stream, switching it over, etc. I wrote a ReleaseBuilder script but of course should actually publish it.
I've also nearly finished the new background, and just need to figure out how to clean the edges of the pic to make it tile properly. Given I'm writing this mail on my phone I will have to mail the proposed image separately, at a later stage.
For better or worse, here it is. Test it with
ActiveWorld color: (InfiniteForm with: (PNGReadWriter
readFromFileNamed: 'my/path/to/blur_ulam_1.png'))
Note the strange vertical white-ish line between the tiles. The image is filled all the way to the edge, but the left and right edges differ sufficiently to create the artifact. I'm not sure how to get rid of it, never having attempted to make a tile-able image before.
Anyone spot anything missing off the swiki release page?
That would be http://wiki.squeak.org/squeak/6188.
frank
Will this do? Done in Gimp in two ticks :-)
Only if you tell me how you did that! I had a brief look over the Gimp's menus (because that's how I made the image in the first place), but nothing leapt out at me. I'm not sure whether I prefer the diamond-ish look or the rectilinear look now, but that's an OK problem to have :).
frank
On 21/11/12 17:03, Frank Shearar wrote:
On 21 November 2012 16:49, Derek O'Connelldoc@doconnel.f9.co.uk wrote:
On 21/11/12 15:53, Frank Shearar wrote:
On 21 November 2012 14:39, Frank Sheararfrank.shearar@gmail.com wrote:
Hi folks,
Apologies for the silence: things have been a bit busy.
Test wise we're looking pretty good. We still have a few red lights but I don't see much point in delaying further: some of the test failures are because of VM sensitivities.
Off the top of my head we/I still need to write a changelog, and otherwise a bunch of infrastructural stuff's all that's needed: a new update stream, switching it over, etc. I wrote a ReleaseBuilder script but of course should actually publish it.
I've also nearly finished the new background, and just need to figure out how to clean the edges of the pic to make it tile properly. Given I'm writing this mail on my phone I will have to mail the proposed image separately, at a later stage.
For better or worse, here it is. Test it with
ActiveWorld color: (InfiniteForm with: (PNGReadWriter
readFromFileNamed: 'my/path/to/blur_ulam_1.png'))
Note the strange vertical white-ish line between the tiles. The image is filled all the way to the edge, but the left and right edges differ sufficiently to create the artifact. I'm not sure how to get rid of it, never having attempted to make a tile-able image before.
Anyone spot anything missing off the swiki release page?
That would be http://wiki.squeak.org/squeak/6188.
frank
Will this do? Done in Gimp in two ticks :-)
Only if you tell me how you did that! I had a brief look over the Gimp's menus (because that's how I made the image in the first place), but nothing leapt out at me. I'm not sure whether I prefer the diamond-ish look or the rectilinear look now, but that's an OK problem to have :).
frank
Gimp v2.6.12, Filters/Map/Make Seamless, and if it you don't have that menu option...
On 21 November 2012 17:10, Derek O'Connell doc@doconnel.f9.co.uk wrote:
On 21/11/12 17:03, Frank Shearar wrote:
On 21 November 2012 16:49, Derek O'Connelldoc@doconnel.f9.co.uk wrote:
On 21/11/12 15:53, Frank Shearar wrote:
On 21 November 2012 14:39, Frank Sheararfrank.shearar@gmail.com wrote:
Hi folks,
Apologies for the silence: things have been a bit busy.
Test wise we're looking pretty good. We still have a few red lights but I don't see much point in delaying further: some of the test failures are because of VM sensitivities.
Off the top of my head we/I still need to write a changelog, and otherwise a bunch of infrastructural stuff's all that's needed: a new update stream, switching it over, etc. I wrote a ReleaseBuilder script but of course should actually publish it.
I've also nearly finished the new background, and just need to figure out how to clean the edges of the pic to make it tile properly. Given I'm writing this mail on my phone I will have to mail the proposed image separately, at a later stage.
For better or worse, here it is. Test it with
ActiveWorld color: (InfiniteForm with: (PNGReadWriter
readFromFileNamed: 'my/path/to/blur_ulam_1.png'))
Note the strange vertical white-ish line between the tiles. The image is filled all the way to the edge, but the left and right edges differ sufficiently to create the artifact. I'm not sure how to get rid of it, never having attempted to make a tile-able image before.
Anyone spot anything missing off the swiki release page?
That would be http://wiki.squeak.org/squeak/6188.
frank
Will this do? Done in Gimp in two ticks :-)
Only if you tell me how you did that! I had a brief look over the Gimp's menus (because that's how I made the image in the first place), but nothing leapt out at me. I'm not sure whether I prefer the diamond-ish look or the rectilinear look now, but that's an OK problem to have :).
frank
Gimp v2.6.12, Filters/Map/Make Seamless, and if it you don't have that menu option...
Nice! Thanks!
frank
It's always fun to try it with Squeak. convert the form to 32 bits and try this:
0 to: self height - 1 do: [ :y | self colorAt: self width-1@y put: ((self colorAt: 0@y) mixed: 0.5 with: (self colorAt: self width-2@y)). self colorAt: 0@y put: ((self colorAt: 2@y) mixed: 0.5 with: (self colorAt: self width-1@y)) ].
Seems to fuzz it up pretty well.
Cheers, Bob
On 11/21/12 10:53 AM, Frank Shearar wrote:
On 21 November 2012 14:39, Frank Shearar frank.shearar@gmail.com wrote:
Hi folks,
Apologies for the silence: things have been a bit busy.
Test wise we're looking pretty good. We still have a few red lights but I don't see much point in delaying further: some of the test failures are because of VM sensitivities.
Off the top of my head we/I still need to write a changelog, and otherwise a bunch of infrastructural stuff's all that's needed: a new update stream, switching it over, etc. I wrote a ReleaseBuilder script but of course should actually publish it.
I've also nearly finished the new background, and just need to figure out how to clean the edges of the pic to make it tile properly. Given I'm writing this mail on my phone I will have to mail the proposed image separately, at a later stage.
For better or worse, here it is. Test it with
ActiveWorld color: (InfiniteForm with: (PNGReadWriter
readFromFileNamed: 'my/path/to/blur_ulam_1.png'))
Note the strange vertical white-ish line between the tiles. The image is filled all the way to the edge, but the left and right edges differ sufficiently to create the artifact. I'm not sure how to get rid of it, never having attempted to make a tile-able image before.
Anyone spot anything missing off the swiki release page?
That would be http://wiki.squeak.org/squeak/6188.
frank
On 21 November 2012 18:22, Bob Arning arning315@comcast.net wrote:
It's always fun to try it with Squeak. convert the form to 32 bits and try this:
0 to: self height - 1 do: [ :y | self colorAt: self width-1@y put: ((self colorAt: 0@y) mixed: 0.5 with: (self colorAt: self width-2@y)). self colorAt: 0@y put: ((self colorAt: 2@y) mixed: 0.5 with: (self colorAt: self width-1@y)) ].
Seems to fuzz it up pretty well.
That's exactly what I'd like to do. The form's an 8 bit ColorForm though, and doesn't like depths > 8. So I thought I'd be extra clever and make a 32 bit form and copy the bits:
| form | form := PNGReadWriter formFromFileNamed: '/home/frank/Downloads/blur_ulam_1.png'. form := (Form extent: form extent depth: 32) bits: form bits. 0 to: form height - 1 do: [ :y | form colorAt: form width-1@y put: ((form colorAt: 0@y) mixed: 0.5 with: (form colorAt: form width-2@y)). form colorAt: 0@y put: ((form colorAt: 2@y) mixed: 0.5 with: (form colorAt: form width-1@y)) asFloat ].
ActiveWorld color: (InfiniteForm with: form)
However, I still can't run the script, because I get an "Error: Bad BitBlt arg (Fraction?). I can't even just wrap-and-resume the Error because, of course, Error is not resumable.
Ideas? What am I missing?
frank
Cheers, Bob
On 11/21/12 10:53 AM, Frank Shearar wrote:
On 21 November 2012 14:39, Frank Shearar frank.shearar@gmail.com wrote:
Hi folks,
Apologies for the silence: things have been a bit busy.
Test wise we're looking pretty good. We still have a few red lights but I don't see much point in delaying further: some of the test failures are because of VM sensitivities.
Off the top of my head we/I still need to write a changelog, and otherwise a bunch of infrastructural stuff's all that's needed: a new update stream, switching it over, etc. I wrote a ReleaseBuilder script but of course should actually publish it.
I've also nearly finished the new background, and just need to figure out how to clean the edges of the pic to make it tile properly. Given I'm writing this mail on my phone I will have to mail the proposed image separately, at a later stage.
For better or worse, here it is. Test it with
ActiveWorld color: (InfiniteForm with: (PNGReadWriter
readFromFileNamed: 'my/path/to/blur_ulam_1.png'))
Note the strange vertical white-ish line between the tiles. The image is filled all the way to the edge, but the left and right edges differ sufficiently to create the artifact. I'm not sure how to get rid of it, never having attempted to make a tile-able image before.
Anyone spot anything missing off the swiki release page?
That would be http://wiki.squeak.org/squeak/6188.
frank
On 2012-11-22, at 14:27, Frank Shearar frank.shearar@gmail.com wrote:
On 21 November 2012 18:22, Bob Arning arning315@comcast.net wrote:
It's always fun to try it with Squeak. convert the form to 32 bits and try this:
0 to: self height - 1 do: [ :y | self colorAt: self width-1@y put: ((self colorAt: 0@y) mixed: 0.5 with: (self colorAt: self width-2@y)). self colorAt: 0@y put: ((self colorAt: 2@y) mixed: 0.5 with: (self colorAt: self width-1@y)) ].
Seems to fuzz it up pretty well.
That's exactly what I'd like to do. The form's an 8 bit ColorForm though, and doesn't like depths > 8. So I thought I'd be extra clever and make a 32 bit form and copy the bits:
| form | form := PNGReadWriter formFromFileNamed: '/home/frank/Downloads/blur_ulam_1.png'. form := (Form extent: form extent depth: 32) bits: form bits.
You can't just use an 8 bit form's bits in a 32 bit form. Use asFormOfDepth:.
- Bert -
0 to: form height - 1 do: [ :y | form colorAt: form width-1@y put: ((form colorAt: 0@y) mixed: 0.5 with: (form colorAt: form width-2@y)). form colorAt: 0@y put: ((form colorAt: 2@y) mixed: 0.5 with: (form colorAt: form width-1@y)) asFloat ].
ActiveWorld color: (InfiniteForm with: form)
However, I still can't run the script, because I get an "Error: Bad BitBlt arg (Fraction?). I can't even just wrap-and-resume the Error because, of course, Error is not resumable.
Ideas? What am I missing?
frank
Cheers, Bob
On 11/21/12 10:53 AM, Frank Shearar wrote:
On 21 November 2012 14:39, Frank Shearar frank.shearar@gmail.com wrote:
Hi folks,
Apologies for the silence: things have been a bit busy.
Test wise we're looking pretty good. We still have a few red lights but I don't see much point in delaying further: some of the test failures are because of VM sensitivities.
Off the top of my head we/I still need to write a changelog, and otherwise a bunch of infrastructural stuff's all that's needed: a new update stream, switching it over, etc. I wrote a ReleaseBuilder script but of course should actually publish it.
I've also nearly finished the new background, and just need to figure out how to clean the edges of the pic to make it tile properly. Given I'm writing this mail on my phone I will have to mail the proposed image separately, at a later stage.
For better or worse, here it is. Test it with
ActiveWorld color: (InfiniteForm with: (PNGReadWriter readFromFileNamed: 'my/path/to/blur_ulam_1.png'))
Note the strange vertical white-ish line between the tiles. The image is filled all the way to the edge, but the left and right edges differ sufficiently to create the artifact. I'm not sure how to get rid of it, never having attempted to make a tile-able image before.
Anyone spot anything missing off the swiki release page?
That would be http://wiki.squeak.org/squeak/6188.
frank
On 22 November 2012 14:27, Bert Freudenberg bert@freudenbergs.de wrote:
On 2012-11-22, at 14:27, Frank Shearar frank.shearar@gmail.com wrote:
On 21 November 2012 18:22, Bob Arning arning315@comcast.net wrote:
It's always fun to try it with Squeak. convert the form to 32 bits and try this:
0 to: self height - 1 do: [ :y | self colorAt: self width-1@y put: ((self colorAt: 0@y) mixed: 0.5 with: (self colorAt: self width-2@y)). self colorAt: 0@y put: ((self colorAt: 2@y) mixed: 0.5 with: (self colorAt: self width-1@y)) ].
Seems to fuzz it up pretty well.
That's exactly what I'd like to do. The form's an 8 bit ColorForm though, and doesn't like depths > 8. So I thought I'd be extra clever and make a 32 bit form and copy the bits:
| form | form := PNGReadWriter formFromFileNamed: '/home/frank/Downloads/blur_ulam_1.png'. form := (Form extent: form extent depth: 32) bits: form bits.
You can't just use an 8 bit form's bits in a 32 bit form. Use asFormOfDepth:.
Ah! That works. OK, so I need just fix the horizontal glitch in a similar manner and I'll be happy.
Thanks, Bert!
frank
- Bert -
0 to: form height - 1 do: [ :y | form colorAt: form width-1@y put: ((form colorAt: 0@y) mixed: 0.5 with: (form colorAt: form width-2@y)). form colorAt: 0@y put: ((form colorAt: 2@y) mixed: 0.5 with: (form colorAt: form width-1@y)) asFloat ].
ActiveWorld color: (InfiniteForm with: form)
However, I still can't run the script, because I get an "Error: Bad BitBlt arg (Fraction?). I can't even just wrap-and-resume the Error because, of course, Error is not resumable.
Ideas? What am I missing?
frank
Cheers, Bob
On 11/21/12 10:53 AM, Frank Shearar wrote:
On 21 November 2012 14:39, Frank Shearar frank.shearar@gmail.com wrote:
Hi folks,
Apologies for the silence: things have been a bit busy.
Test wise we're looking pretty good. We still have a few red lights but I don't see much point in delaying further: some of the test failures are because of VM sensitivities.
Off the top of my head we/I still need to write a changelog, and otherwise a bunch of infrastructural stuff's all that's needed: a new update stream, switching it over, etc. I wrote a ReleaseBuilder script but of course should actually publish it.
I've also nearly finished the new background, and just need to figure out how to clean the edges of the pic to make it tile properly. Given I'm writing this mail on my phone I will have to mail the proposed image separately, at a later stage.
For better or worse, here it is. Test it with
ActiveWorld color: (InfiniteForm with: (PNGReadWriter readFromFileNamed: 'my/path/to/blur_ulam_1.png'))
Note the strange vertical white-ish line between the tiles. The image is filled all the way to the edge, but the left and right edges differ sufficiently to create the artifact. I'm not sure how to get rid of it, never having attempted to make a tile-able image before.
Anyone spot anything missing off the swiki release page?
That would be http://wiki.squeak.org/squeak/6188.
frank
You could also just swap some of the pixels:
| w h f | f := (PNGReadWriter formFromFileNamed: '/Users/bob/squeak/Squeak4.3/blur_ulam_1.png') asFormOfDepth: 32. w := f width - 1. h := f height - 1. 0 to: h by: 2 do: [ :y || c | c := f colorAt: 0@y. f colorAt: 0@y put: (f colorAt: w@y). f colorAt: w@y put: c. ]. 0 to: w by: 2 do: [ :x || c | c := f colorAt: x@0. f colorAt: x@0 put: (f colorAt: x@h). f colorAt: x@h put: c. ]. ActiveWorld color: (InfiniteForm with: f)
Cheers, Bob
On 11/22/12 10:14 AM, Frank Shearar wrote:
On 22 November 2012 14:27, Bert Freudenberg bert@freudenbergs.de wrote:
On 2012-11-22, at 14:27, Frank Shearar frank.shearar@gmail.com wrote:
On 21 November 2012 18:22, Bob Arning arning315@comcast.net wrote:
It's always fun to try it with Squeak. convert the form to 32 bits and try this:
0 to: self height - 1 do: [ :y | self colorAt: self width-1@y put: ((self colorAt: 0@y) mixed: 0.5 with: (self colorAt: self width-2@y)). self colorAt: 0@y put: ((self colorAt: 2@y) mixed: 0.5 with: (self colorAt: self width-1@y)) ].
Seems to fuzz it up pretty well.
That's exactly what I'd like to do. The form's an 8 bit ColorForm though, and doesn't like depths > 8. So I thought I'd be extra clever and make a 32 bit form and copy the bits:
| form | form := PNGReadWriter formFromFileNamed: '/home/frank/Downloads/blur_ulam_1.png'. form := (Form extent: form extent depth: 32) bits: form bits.
You can't just use an 8 bit form's bits in a 32 bit form. Use asFormOfDepth:.
Ah! That works. OK, so I need just fix the horizontal glitch in a similar manner and I'll be happy.
Thanks, Bert!
frank
- Bert -
0 to: form height - 1 do: [ :y | form colorAt: form width-1@y put: ((form colorAt: 0@y) mixed: 0.5 with: (form colorAt: form width-2@y)). form colorAt: 0@y put: ((form colorAt: 2@y) mixed: 0.5 with: (form colorAt: form width-1@y)) asFloat ].
ActiveWorld color: (InfiniteForm with: form)
However, I still can't run the script, because I get an "Error: Bad BitBlt arg (Fraction?). I can't even just wrap-and-resume the Error because, of course, Error is not resumable.
Ideas? What am I missing?
frank
Cheers, Bob
On 11/21/12 10:53 AM, Frank Shearar wrote:
On 21 November 2012 14:39, Frank Shearar frank.shearar@gmail.com wrote:
Hi folks,
Apologies for the silence: things have been a bit busy.
Test wise we're looking pretty good. We still have a few red lights but I don't see much point in delaying further: some of the test failures are because of VM sensitivities.
Off the top of my head we/I still need to write a changelog, and otherwise a bunch of infrastructural stuff's all that's needed: a new update stream, switching it over, etc. I wrote a ReleaseBuilder script but of course should actually publish it.
I've also nearly finished the new background, and just need to figure out how to clean the edges of the pic to make it tile properly. Given I'm writing this mail on my phone I will have to mail the proposed image separately, at a later stage.
For better or worse, here it is. Test it with
ActiveWorld color: (InfiniteForm with: (PNGReadWriter
readFromFileNamed: 'my/path/to/blur_ulam_1.png'))
Note the strange vertical white-ish line between the tiles. The image is filled all the way to the edge, but the left and right edges differ sufficiently to create the artifact. I'm not sure how to get rid of it, never having attempted to make a tile-able image before.
Anyone spot anything missing off the swiki release page?
That would be http://wiki.squeak.org/squeak/6188.
frank
This gives a slightly (as in "I can see it with a MagnifierMorph but otherwise not) cleaner horizontal join, together with your original nice vertical join:
| form | form := (PNGReadWriter formFromFileNamed: '/home/frank/Documents/squeak/blur_ulam_1.png') asFormOfDepth: 32.
0 to: form height - 1 do: [ :y | form colorAt: form width-1@y put: ((form colorAt: 0@y) mixed: 0.5 with: (form colorAt: form width-2@y)). form colorAt: 0@y put: ((form colorAt: 2@y) mixed: 0.5 with: (form colorAt: form width-1@y)) ].
0 to: form width do: [:x | | top bottom | top := form colorAt: x@0. bottom := form colorAt: x@(form height - 1). form colorAt: x@0 put: (top mixed: 0.5 with: bottom). form colorAt: x@(form height - 1) put: (top mixed: 0.5 with: bottom)].
ActiveWorld color: (InfiniteForm with: form).
"MagnifierMorph newRound openInWorld"
On 22 November 2012 17:25, Bob Arning arning315@comcast.net wrote:
You could also just swap some of the pixels:
| w h f | f := (PNGReadWriter formFromFileNamed: '/Users/bob/squeak/Squeak4.3/blur_ulam_1.png') asFormOfDepth: 32. w := f width - 1. h := f height - 1. 0 to: h by: 2 do: [ :y || c | c := f colorAt: 0@y. f colorAt: 0@y put: (f colorAt: w@y). f colorAt: w@y put: c. ]. 0 to: w by: 2 do: [ :x || c | c := f colorAt: x@0. f colorAt: x@0 put: (f colorAt: x@h). f colorAt: x@h put: c. ]. ActiveWorld color: (InfiniteForm with: f)
Cheers, Bob
On 11/22/12 10:14 AM, Frank Shearar wrote:
On 22 November 2012 14:27, Bert Freudenberg bert@freudenbergs.de wrote:
On 2012-11-22, at 14:27, Frank Shearar frank.shearar@gmail.com wrote:
On 21 November 2012 18:22, Bob Arning arning315@comcast.net wrote:
It's always fun to try it with Squeak. convert the form to 32 bits and try this:
0 to: self height - 1 do: [ :y | self colorAt: self width-1@y put: ((self colorAt: 0@y) mixed: 0.5 with: (self colorAt: self width-2@y)). self colorAt: 0@y put: ((self colorAt: 2@y) mixed: 0.5 with: (self colorAt: self width-1@y)) ].
Seems to fuzz it up pretty well.
That's exactly what I'd like to do. The form's an 8 bit ColorForm though, and doesn't like depths > 8. So I thought I'd be extra clever and make a 32 bit form and copy the bits:
| form | form := PNGReadWriter formFromFileNamed: '/home/frank/Downloads/blur_ulam_1.png'. form := (Form extent: form extent depth: 32) bits: form bits.
You can't just use an 8 bit form's bits in a 32 bit form. Use asFormOfDepth:.
Ah! That works. OK, so I need just fix the horizontal glitch in a similar manner and I'll be happy.
Thanks, Bert!
frank
- Bert -
0 to: form height - 1 do: [ :y | form colorAt: form width-1@y put: ((form colorAt: 0@y) mixed: 0.5 with: (form colorAt: form width-2@y)). form colorAt: 0@y put: ((form colorAt: 2@y) mixed: 0.5 with: (form colorAt: form width-1@y)) asFloat ].
ActiveWorld color: (InfiniteForm with: form)
However, I still can't run the script, because I get an "Error: Bad BitBlt arg (Fraction?). I can't even just wrap-and-resume the Error because, of course, Error is not resumable.
Ideas? What am I missing?
frank
Cheers, Bob
On 11/21/12 10:53 AM, Frank Shearar wrote:
On 21 November 2012 14:39, Frank Shearar frank.shearar@gmail.com wrote:
Hi folks,
Apologies for the silence: things have been a bit busy.
Test wise we're looking pretty good. We still have a few red lights but I don't see much point in delaying further: some of the test failures are because of VM sensitivities.
Off the top of my head we/I still need to write a changelog, and otherwise a bunch of infrastructural stuff's all that's needed: a new update stream, switching it over, etc. I wrote a ReleaseBuilder script but of course should actually publish it.
I've also nearly finished the new background, and just need to figure out how to clean the edges of the pic to make it tile properly. Given I'm writing this mail on my phone I will have to mail the proposed image separately, at a later stage.
For better or worse, here it is. Test it with
ActiveWorld color: (InfiniteForm with: (PNGReadWriter readFromFileNamed: 'my/path/to/blur_ulam_1.png'))
Note the strange vertical white-ish line between the tiles. The image is filled all the way to the edge, but the left and right edges differ sufficiently to create the artifact. I'm not sure how to get rid of it, never having attempted to make a tile-able image before.
Anyone spot anything missing off the swiki release page?
That would be http://wiki.squeak.org/squeak/6188.
frank
To tell the truth, I needed to get my eyes at a very precise distance from the screen to even see the horizontal artifact. I guess age is another kind of filter. ;-)
Cheers, Bob
On 11/22/12 1:44 PM, Frank Shearar wrote:
This gives a slightly (as in "I can see it with a MagnifierMorph but otherwise not) cleaner horizontal join, together with your original nice vertical join:
For Windows the fix I made make the test error go away. I'm not sure it is a good fix.
Karl
On Wed, Nov 21, 2012 at 3:39 PM, Frank Shearar frank.shearar@gmail.com wrote:
Hi folks,
Apologies for the silence: things have been a bit busy.
Test wise we're looking pretty good. We still have a few red lights but I don't see much point in delaying further: some of the test failures are because of VM sensitivities.
Off the top of my head we/I still need to write a changelog, and otherwise a bunch of infrastructural stuff's all that's needed: a new update stream, switching it over, etc. I wrote a ReleaseBuilder script but of course should actually publish it.
I've also nearly finished the new background, and just need to figure out how to clean the edges of the pic to make it tile properly. Given I'm writing this mail on my phone I will have to mail the proposed image separately, at a later stage.
Anyone spot anything missing off the swiki release page?
frank
squeak-dev@lists.squeakfoundation.org