For authenticated file transfers, is there any reason to use ftp instead of the ssh file transfer protocol (sftp)?
Depends on how they're authenticated. If your customer has a shell account on your machine, you're right. But some ftp servers that authenticate against, say, ldap or a database. Keeps your customers out of/etc/passwd.
You could certainly do this for a few files with http. But when there are ~2000 files totaling ~100 GB, and the customer is of the old school who probably doesn't know (or care) what torrent is, ftp is the way to go.
...but I've searched high-res sky images in the past without finding anything systematic except some specific projects such the Sloan Sky survey (which are just coordinates) or the odd marketing Hubble shot.
If that's all you found, you didn't look hard enough. Sloan serves imaging and spectral data, and all of Hubble's science data (for example) has been available from three different data centers since 1992. (This is data we're talking about, not pretty pictures.) In fact, all NASA-funded missions are required to archive their data, and NSF is (finally) getting into the act. I don't know what ESA requires, but I know they're building a large archive. And just about every large ground-based project in development has a significant archival component.
I'd say your previous employer's attitude ("our funding, our data") is the exception nowadays. Even privately funded projects are looking at archives, if only to connect to the VO.
Not sure what's meant exactly by it being the "only collection to cover both hemispheres". The Digitized Sky Survey covers the whole sky and it's been online for 12 years.
You're being naive, my friend. Boards of Directors aren't objective guardians of shareholders' interest. They are more often friends of the CEO, or his kids' piano teacher, or other CEOs who have no intention of messing with the compensation gravy train. So they heap on the options to make it look like the CEO has a "stake" in the company's success, when in fact he has nothing at risk and can reward himself handsomely for mediocre performance. Between 1990 and 2003, Michael Eisner took in $800 million as CEO of Disney, while the shareholders did no better than they would have in Treasurys.
This is well explained in Roger Lowenstein's Origins of the Crash. Read it, if you haven't completely closed your mind.
What happened? I remember when we were told that aiming Hubble at the Moon or the Earth would destroy it's sensitive instruments.
Hubble can do short images of the moon with no problem, aside from the challenge of guiding. It does images of the earth all the time. These are called earth calibrations and they serve as the basis of flat fields with which HST images are calibrated. You can't see anything in them, though, because the earth is too close to focus on, and the telescope is moving at ~300 miles/min, so the images are just blurry streaks across cloud tops. That's why they make good flat fields.
Not long after launch, HST did some "imaging" of the sun. The idea was to point the telescope 180 degrees away from the sun while using a small backwards-pointing light collector on the original WF/PC to pre-flood the CCD with solar ultraviolet. It never got used , though. HST Proposal 1478: WF/PC UV FLOOD GUIDING TECHNIQUE VERIFICATION, if you're interested. Here's an example image.
So the only major solar system object that HST has never imaged-besides the objects we don't know about-is Mercury. It's too close to the sun. The aperture door will close if we try to point there.
Our problem was that we have database searches that return potentially hungous numbers of records, which take up memory, which isn't given back to the system. Having lots of giant Apache processes around drove us to make the Apache process die when it was done. I forget the name of the PHP command that does this, but I recall from my testing that it didn't work with Apache 2. Does it now? or is memory usage handled differently now?
(Yes, I know the best solution would be to page the results, but that would require significant retooling of the application - long story - and I wanted the upgrade to be a simple drop-in replacement with no code changes. And we're short on resources, so proxying such requests would not buy us much.)
The WFC detector, called ACS/WFC, employs a mosaic of two 2048 × 4096 Scientific Imaging Technologies (SITe) CCDs, with ~0.05 arcsecond pixels, covering a nominal 202 × 202 arcsecond field of view (FOV), and a spectral response from~3700 to 11,000 Å.
The HRC detector, called ACS/HRC, is a 1024 × 1024 SITe CCD, with ~0.028 × 0.025 arcsecond pixels, covering a nominal 29 × 26 arcsecond field of view, and spectral response from ~2000 to 11,000 Å.
The SBC detector, called the ACS/SBC, is a solar-blind CsI Multi-Anode Microchannel Array (MAMA), with 1024 × 1024 ~0.034 × 0.030 arcsecond pixels, and a nominal 35 × 31 arcsecond FOV, with far-UV spectral response from 1150 to 1700Å.
Additionally, these images are usually drizzled and mosaicked to create larger images. The GOODSsections, for example, are each 8192x8192, and the overall size of the field would have been 32768x40960, but then you would need a 64-bit machine to work with it (which is why it was broken into tiles).
Oh, and keep *this* in mind: These detectors are space rated. A consumer-grade digi-cam is not.
...anybody have a clue exactly what happens when a telescope dies?
Usually it runs out of money. IUE was supposed to go for five years; instead it went along for 18. It was ailing some toward the end, but ultimately it was funding that killed it.
This is something I've noticed with Hubble, except that we who deal with the technical side of it don't wax nearly as emotional about it as the astronomers who work with the data.
Grab a copy of the Summary Tables for the US budget and take a look at Table S-3 on page 5. It shows the budgets for various agencies and how they've grown or shrunk since 2001.
Slashdot Mirror
Depends on how they're authenticated. If your customer has a shell account on your machine, you're right. But some ftp servers that authenticate against, say, ldap or a database. Keeps your customers out of
You could certainly do this for a few files with http. But when there are ~2000 files totaling ~100 GB, and the customer is of the old school who probably doesn't know (or care) what torrent is, ftp is the way to go.
The DSS is a collection of plates that covers both hemispheres.
If that's all you found, you didn't look hard enough. Sloan serves imaging and spectral data, and all of Hubble's science data (for example) has been available from three different data centers since 1992. (This is data we're talking about, not pretty pictures.) In fact, all NASA-funded missions are required to archive their data, and NSF is (finally) getting into the act. I don't know what ESA requires, but I know they're building a large archive. And just about every large ground-based project in development has a significant archival component.
I'd say your previous employer's attitude ("our funding, our data") is the exception nowadays. Even privately funded projects are looking at archives, if only to connect to the VO.
Not sure what's meant exactly by it being the "only collection to cover both hemispheres". The Digitized Sky Survey covers the whole sky and it's been online for 12 years.
You're being naive, my friend. Boards of Directors aren't objective guardians of shareholders' interest. They are more often friends of the CEO, or his kids' piano teacher, or other CEOs who have no intention of messing with the compensation gravy train. So they heap on the options to make it look like the CEO has a "stake" in the company's success, when in fact he has nothing at risk and can reward himself handsomely for mediocre performance. Between 1990 and 2003, Michael Eisner took in $800 million as CEO of Disney, while the shareholders did no better than they would have in Treasurys.
This is well explained in Roger Lowenstein's Origins of the Crash. Read it, if you haven't completely closed your mind.
Jon Carrol's column in SFGate has a little comparison.
What happened? I remember when we were told that aiming Hubble at the Moon or the Earth would destroy it's sensitive instruments.
Hubble can do short images of the moon with no problem, aside from the challenge of guiding. It does images of the earth all the time. These are called earth calibrations and they serve as the basis of flat fields with which HST images are calibrated. You can't see anything in them, though, because the earth is too close to focus on, and the telescope is moving at ~300 miles/min, so the images are just blurry streaks across cloud tops. That's why they make good flat fields.
Not long after launch, HST did some "imaging" of the sun. The idea was to point the telescope 180 degrees away from the sun while using a small backwards-pointing light collector on the original WF/PC to pre-flood the CCD with solar ultraviolet. It never got used , though. HST Proposal 1478: WF/PC UV FLOOD GUIDING TECHNIQUE VERIFICATION, if you're interested. Here's an example image.
So the only major solar system object that HST has never imaged-besides the objects we don't know about-is Mercury. It's too close to the sun. The aperture door will close if we try to point there.
That should apply to everybody, not just the poor. This is a good chance for a lot of people to learn to live without that damn box.
Since Sunday it's been AN HOUR AFTER NOW.
Actually, a kid can be a pretty good gadget in itself.
"Take the personal out of computing, and most companies would grind to a halt."
"Now, let's talk about Web Services!"
Our problem was that we have database searches that return potentially hungous numbers of records, which take up memory, which isn't given back to the system. Having lots of giant Apache processes around drove us to make the Apache process die when it was done. I forget the name of the PHP command that does this, but I recall from my testing that it didn't work with Apache 2. Does it now? or is memory usage handled differently now?
(Yes, I know the best solution would be to page the results, but that would require significant retooling of the application - long story - and I wanted the upgrade to be a simple drop-in replacement with no code changes. And we're short on resources, so proxying such requests would not buy us much.)
Meanwhile, here's what the ACS has for detectors:
- The WFC detector, called ACS/WFC, employs a mosaic of two 2048 × 4096 Scientific Imaging Technologies (SITe) CCDs, with ~0.05 arcsecond pixels, covering a nominal 202 × 202 arcsecond field of view (FOV), and a spectral response from~3700 to 11,000 Å.
- The HRC detector, called ACS/HRC, is a 1024 × 1024 SITe CCD, with ~0.028 × 0.025 arcsecond pixels, covering a nominal 29 × 26 arcsecond field of view, and spectral response from ~2000 to 11,000 Å.
- The SBC detector, called the ACS/SBC, is a solar-blind CsI Multi-Anode Microchannel Array (MAMA), with 1024 × 1024 ~0.034 × 0.030 arcsecond pixels, and a nominal 35 × 31 arcsecond FOV, with far-UV spectral response from 1150 to 1700Å.
Additionally, these images are usually drizzled and mosaicked to create larger images. The GOODS sections, for example, are each 8192x8192, and the overall size of the field would have been 32768x40960, but then you would need a 64-bit machine to work with it (which is why it was broken into tiles).Oh, and keep *this* in mind: These detectors are space rated. A consumer-grade digi-cam is not.
Finaally, some saanity haas been restored to the paatent debaate. Congraatulaations to our Dutch friends for their couraageous step forwaard.
One problem with sending humans is that you have to bring them down after a week. A robot can stay up for as long as it takes to get the job done.
"Sneed's Feed and Seed" should have "formerly Chuck's" appended.
Not that I watch the show that closely or anything...
...anybody have a clue exactly what happens when a telescope dies?
Usually it runs out of money. IUE was supposed to go for five years; instead it went along for 18. It was ailing some toward the end, but ultimately it was funding that killed it.
This is something I've noticed with Hubble, except that we who deal with the technical side of it don't wax nearly as emotional about it as the astronomers who work with the data.
That's one packet I wouldn't care to sniff.
What's significant is that USA Today noticed.
But when a "study" like this comes out, stating the obvious in "OMFG the sky is falling!" terms, you should follow the money.
Indeed. Just two more researchers in the pockets of the powerful Shower Curtain Industry Association of America (SCIAA).
Time Europe recently had a A article about European scientists emmigrating to the US because it was easier to do science there than in Europe (less bureaucracy in the US, though we're catching up).
Grab a copy of the Summary Tables for the US budget and take a look at Table S-3 on page 5. It shows the budgets for various agencies and how they've grown or shrunk since 2001.
Some projects seem to use "-ux" or "-ix" at the end for this purpose, e.g. Knoppix.
And to a lesser extent, Winn-Dixie.
Hubble does in fact take images of the earth quite often. They're used as the basis of flat fields, which are used to calibrate the science data.