"Do you really think it is in Apple or Samsung's interest to add to a customer's carrier lock-in?"
Even if it isn't in their interest there's always the likelihood of unintended consequences.
16426 posts • joined 16 Jun 2014
"There seem to be a large number of comments here asking why the google cars (how many cars do they have in total?) have been in accidents in the 6 years they have been running."
Those comments are largely in response to my original comment which started out with the word "If". I've no idea if the frequency is excessive. But if they are then it's reasonable to ask why.
Much the same applies to stretches of roads with particularly high accident rates. An example would be one road near me. It has:
- A cross-roads with another main road with particularly bad sight lines for both roads.
- A junction on a bend where the minor road is a tangent to the main road. The main road has a 50 speed limit but the side road has a national speed-limit sign which catches headlights of drivers on the main road at night & looks as if it's the main road continuing in a straight line - even when you know it's a side road. I know there's been a head-on fatal collision on that bend.
- Two more minor cross roads both with bad sight lines although one has little traffic & has been ameliorated with a mirror. Drivers on the main road approaching the latter are distracted by a flashing slow-down sign which would probably be triggered by a snail with a sufficiently large radar cross-section.
- Some not very well cambered bends.
It's plain that the major junction needs to be re-configured - the current staggered form just exposes crossing traffic for long than a straight crossing would do. It needs a roundabout or possibly a pair. The bend with the speed-limit needs reflectors to show the outline at night. The worst of the the other cross-roads needs traffic lights.
What do they get? Signs advising drivers it's a high risk road and playing about with speed limits. You do not improve elevated risks on the road without asking why they're elevated and then tackling the specific issues. Don't blame the drivers if the road is the anomaly.
In the case of Google cars IF there is an anomalous risk then one does need to ask why.
If the accidents are happening to Googlemoblies significantly more often than to conventionally driven cars then it does raise the question of what might be the explanation. What is different about the behaviour of these vehicles which contradicts the following drivers' expectations?
I hear the argument about naming and shaming discouraging self-reporting. But if self-reporting is mandatory encouragement and discouragement are irrelevant.
What is relevant is that the police raid is unlikely to have been invisible to the neighbours so in effect an innocent party has been been named and shamed. Any effective apology would have to have been a public one in which case the authority responsible must have already outed themselves - and if they didn't they deserved to have been held to public account here.
In my view it's yet another argument for requiring judicial warrants with mandatory reporting of outcomes back to the granter. The possibility of having to report back to a magistrate or judge that they have issued a warrant against the wrong person should concentrate the mind.
Hmm. I thought I'd posted one comment here earlier but don't see it.
I installed an earlier build on a little Atom test box, partly to check on being able to set up FreeBSD dual boot.
It couldn't see either my printer (HP 3020 on a JetDirect box) or my NAS. Initially I couldn't even set the network mask so it could see their IP addresses. It could see my router & saw that it could be a media server (it shows it twice, once for each function) and it could also see my wireless AP. Updates allowed setting of network mask but no joy with printer or NAS.
One positive note was that it would allow me to install a lot of ancient Windows applications (I only have ancient versions of Windows stuff). There's a compatibility dialog to allow it to install as if it was one of several various earlier versions of Windows. This included Lotus Smartsuite (I said ancient) and Office 97 from a competitive upgrade disk (remember those?). It looks as if a lot of legacy programs might run.
I tried installing the current build from scratch. Same old installer bully-boy approach - blows away the existing boot loader but doesn't bother to check for other OSes.
Slow, slow, slow, slow, SSSSSLLLLLOOOOOWWWWW. Not only could I have installed Linux in a fraction of the time, I could have installed Linux, FreeBSD and SCO one after the other - and IIRC SCO was a really long-winded install.
It still doesn't find anything extra on the network automatically. I managed to set the network drive up going through the old settings window which is accessible as a sort of footnote via the new flat version. Manual addition of the printer got as far as driver selection - it didn't have the driver but offered to look for it via Windows Update which was also slow etc & I knocked it on the head after a while. Eventually I downloaded the 8.1 driver from HP and that worked. I haven't tried using the scanner.
The tile part of the start menu can be got rid of. Just unpin each tile in turn and then resize the window. The rest of the menu is still cumbersome.
It looks as if it can be arm-wrestled into working but I won't be keeping it. For the limited amount of stuff that I occasionally have to use Windows for it'll be either the Win7 VM or even the W2K VM.
I installed a previous build on a little Atom box, mostly to see if installing FreeBSD would detect & dual boot another OS (it would). Even after several updates it couldn't find my printer & NAS box. So this evening I decided to install the latest version.
Same bully-boy Windows installer attitude - just blow away the existing boot sector. Does it now find the printer & NAS box? How would I know? As of a couple of minutes ago it was still running the post-install setup. I could have installed Linux in a fraction of the time - no, I could have installed Linux, FreeBSD and SCO, one after the other, in less time. And SCO, as I remember it, was a long-winded installation.
"the LHB occurred between 4.1 and 3.8 billion years ago, the earliest evidence we have for life on Earth is from around 3.7 billion years ago."
If life had developed prior to the LHB the evidence for it might not have survived - indeed life might not have survived - so it's not surprising that the first evidence comes later.
"current theories, and experiments conducted since the '70s suggest that it is rather difficult to get life going on its own"
It's still difficult to get going even if you suppose it happened elsewhere. I've never found this hypothesis appealing - it smacks of trying to avoid a difficult problem by turning your back on it. I suppose the advantage of the hypothesis is that you can allow a much longer time-span for it to have happened. But as a scientific hypothesis it has the disadvantage of being difficult to falsify - look at a comet & find nothing you have a choice of saying "wrong comet" or "wrong type of comet".
"For applications that don't, it can take anywhere from hours to years to solve the roadblocks."
And that doesn't even include company politics. Sequent box running the entire logistics business going out of support at year end. Se well in advance IT planned to migrate to Sun over the Christmas/New Year break when the business was closed for holidays. Told definitely not. Eventually it transpired that the owners had arranged for the books to be gone over to value the business for it to be sold.
As a Unixer with experience dating back to 7th edition days I'd obviously recommend a Unix-like platform wherever feasible. But Trevor's right: you can't make a cross-platform move like that in a month, especially if you've no existing Unix experience. OTOH I'd have thought that a month is cutting it fine even for moving a small estate from one Windows platform to another. Apart from the acquisition of any new hardware you have to allow for testing and plan a good time for the migration. Realistically you have to allow for the possibility that testing will reveal some problems there may be no suitable time for the move within the month, especially if you have to work round some problem discovered in testing.
"The trouble with backs is that spines were designed as beams for four-footed animals. Walking upright we have turned them into columns and the stresses are completely different."
No. They evolved as beams for four-footed animals. Then they evolved into columns. Evolution is effective at optimising things but it can only achieve local maxima. So what we have is probably the best column that can be adapted from an articulated beam but probably not the best column that could have been evolved directly from a notochord and certainly not the best column that could have been designed. And hence, as you say, the backaches & other problems.
"The last time I was given XML from a client - ummm, this year - it wasn't even proper XML."
The way we set things up was that the XML schema was agreed with the client (i.e. my client's client, the main project contractor). I'm not sure whether it was part of the actual contract but every new product or product change was documented in a version controlled spec and in a DTD or schema (which I usually maintained). Nothing went live until we had test data from the client validated against the current schema, processed and the sample product signed off by the main contractor's client. In production any file received which was not a well-formed XML document would be refused. This happened from time to time because the sometimes the latest devs at the other end hadn't grasped the use of entities to handle certain characters. As the devs rotated when their visas ran out I occasionally had to do a bit of education...
We didn't validate the whole XML document but validated the individual fragments representing an order printed document. IIRC we had an arrangement to simply discard and report a particular order that failed validation rather than bounce several hundred good ones.
Although it's fashionable to decry XML as over-engineered it came with a selection of tools to do the heavy lifting and if you made proper use of them it was vastly better than having a system gamely soldier on and do the wrong thing or fall flat on its face when encountering bad data. I can't comment on JSON as I've never used it; does it have the same support for data integrity?
Sometimes the bodges are mandated by the client.
Client ran a digital print service. They got flat files, usually CSV to print. The printers were driven by a package which took in flat files with one line per field in the document together with a formatting in file which told where & how the field was printed. The normal work-flow was to store the contents of the incoming files in a relational database & then pull the data out in the field-per-line format. Having the stuff in the database helped manage batching, remakes etc & also made the conversion from one flat file format to another fairly transparent. The normal IT work-flow had been to write a system for this more or less from scratch for every contract. I'll draw a veil over the contract where the data came via EDI...
Along came a contract which needed this new-fangled XML stuff. The document was way more complex than the usual stuff and flat files wouldn't have handled it. I put myself up to handle the XML end & did some training on the subject. The obvious route was take the original XML & apply XSLT to convert it into the field-per-line format. To handle the usual work-flow requirements the incoming XML could be split into fragments, one per printed document, stored as text elements in the database and reassembled for a batch job. Client said 'No'.
They wanted the XML taken apart and stored in relational form just like all the others except this time it would require a whole hierarchy of tables and it quickly became clear that for performance reasons surrogate keys would have to be used to tie stuff together. I ended up with XSL to convert the XML to SQL with a series of macros to act as place-holders for the keys and a macro-processor to handle the tying together. Inevitably more sections were added to the document format and hence to the XML over the life of the contract. The database design was tied to the document structure and chunks of the code were tied to the schema so the client had committed themselves to changing both at intervals through the life of the contract.
"For example, how can you stop someone being influenced by a lobbyist who also happens to be their spouse or immediate relative?"
If their areas of concern are different there's no problem. But you should require them to declare a conflict of interest and one of them to resile. But maybe I'm old-fashioned.
"Does this imply that you can't tell foreign software companies about security holes you have found in their products?"
AFAICS, yes. It would also be illegal for any criminal to make use of the same holes should they discover them. Smart, very smart. Aren't we lucky we have such smart people looking after us?
"British TV companies now spend less creating original material; scrapping regulations means you get low quality programming; and people are losing the habit of watching live TV."
Let's rearrange that sequence:
1. scrapping regulations means you get low quality programming;
2. British TV companies now spend less creating original material;
3. people are losing the habit of watching live TV.
1 leads to 2, 2 leads to 3 & a feedback loop from 3 to 2 makes the situation a runaway race to the bottom.
"The one thing you have to understand, if you really want to know why things like this happen, is that the ONE overriding concern of everyone involved on the government side of this, is THEY MUST NOT BE ABLE TO HOLD ME RESPONSIBLE."
Quick fix. The Treasury doesn't release funds for any project above £x without having the name of an individual who is held responsible. If cost overruns take a lower cost project up to £x no additional funds are released without their having the name of an individual who is held responsible. For existing projects over £x no further tranches are released without having the name of an individual who is held responsible. And require evidence that the named individuals actually have the clout to exercise that responsibility.
The Treasury hold the purse strings. They can lay down such conditions if they have the will-power to do so. It might cause ructions elsewhere but that's elsewhere's problem.
"taskbar-like panels with applets, a start menu and system tray"
This misses out the most significant part: the ability to put files, folders*, links or applications on the desktop. The current fashion in "user experience design" is that you must follow the designer's workflow whether it's relevant to you or not.
Stuff designing my user experience, just provide a versatile user interface & I'll design my own experience.
No ship allowed to dock in Dover unless all vehicles aboard have been searched in port in Calais after loading. Of course this allow far fewer ships to sale from Calais so most of the traffic would be diverted to Belgium. How long would it take the Frogs to realise that the value to Calais of the cross-channel trade is worth beefing up their security?
If the scope extends to pre-schoolers the old Ladybird Peter & Jane books have a lot going for them. Back in the day they got my son reading within weeks aged about 4. The content may need updating (autre temps, autre mores). The crucial aspect of it was that learning involved three elements, the child, the material & a parent (or other reader) providing one-to-one support. The books provided well-paced material so that the parent didn't have to have skills in teaching reading.
Biting the hand that feeds IT © 1998–2019