Sunday, 18 February 2018

Integrity v8 engine under development


The beta of link checker Integrity v8 is imminent, with Integrity Plus following closely behind and then website scrutineer Scrutiny.

Obviously we've been looking forward to version 8 so that we can say 'V8 engine'. The biggest changes in the upcoming versions of Integrity and Scrutiny are in the engine rather than interface, so there's not a huge amount to see on the surface but there is a lot to say about it.

Changes to the way that the data is structured in memory

For 11 years, the way that the data has been structured in Integrity and its siblings has meant that for certain views, tables had to be built following the scan. And rebuilt when certain actions were taken (rechecking, marking as fixed). Not a big deal with small sites, but heavy on time and resources for very large sites. It also meant that certain information wasn't readily available (number of internal backlinks, for example) and had to be looked up or calculated either on the fly or as a background task after the scan. And I have to admit to some cases of instability, unresponsiveness or unexpected side effects caused over the years by that stuff going on in the background.

So we've invested a lot of time in something that isn't a killer new feature, but will make things run more smoothly, efficiently and reliably. Initial informal tests show improvements in speed and efficiency (memory use).

More information about your links

There have been a number of requests to add more information about the properties of a link , eg hreflang and more of the 'rel' attributes (it has been displaying the 'nofollow' status but not much else). HTML5 allows loads of allowable values in the 'rel' and if the one you want doesn't have its own yes/no column in the new apps then you can view the entire rel attribute. You can switch these columns on in any of the link tables as well as the one in the link inspector.


The information displayed in Scrutiny's SEO table will include more data about the pages, notably og: meta data which is important to a lot of users.

More information about redirects

Redirect information wasn't stored permanently, only the start and end of a redirect chain (the majority are a single step. But if there is more than one redirect, you'll want to know all the in-between details)


Better control over volume of requests

It's becoming increasingly necessary to apply some limits on the rate of the requests that Integrity / Scrutiny make.

Your server may keep responding no matter how hard you hit it. That's great; turn up those threads and watch Integrity / Scrutiny go through your site like a dose of salts.

But I'm seeing more cases of servers objecting to rapid requests from the same source. And the reason isn't always clear. They may return 429 (which means 'too many requests') or some other less obvious error response. Or they may just stop responding, perhaps a defensive measure.

If that's your site, you'll no longer have to turn down the threads and use trial and error with the delay field. You'll be able to simply enter a maximum number of requests per minute, while still using multiple threads for efficiency.


Why wasn't I informed about Integrity version 7?

Because it doesn't exist; Integrity will skip a version. v7 of Scrutiny was mostly about the interface. Integrity continued to use the classic interface and so remained v6. Integrity, Integrity Plus and Scrutiny (and a new app, Integrity Pro, you heard it here first) will use the v8 engine when it's ready and so for consistency all apps will be called v8.

Charge for an upgrade?  Price increase?

So in summary, as far as Scrutiny v8 and Integrity v8 are concerned, the changes won't be immediately visible but they'll do what they do more efficiently and with more information available.   They'll provide more information more efficiently in a smaller memory footprint.

There are no immediate plans for price increases, or for upgrade fees. Prices are of course always under review, but any future increases will be for new users, upgrading to 8 from Scrutiny 7 or Integrity 6 will be free.

Sunday, 4 February 2018

iPhone Battery & Power Repair (UK)

I didn't find this as easy to find as I'd hoped, so I'm posting the link here.


It applies to iPhone 6 and later. The £25 offer covers the battery replacement *if they test it and decide that the battery needs replacing*. (£79 for 'other eligible models' but no word on which models.) It may be free if you're under warranty or AppleCare+. You can send it to them or take it into an Apple store.

Read more and start a request:
https://support.apple.com/en-gb/iphone/repair/battery-power

Thursday, 1 February 2018

MP Matt Hancock releases app called Matt Hancock MP

This story is the best laugh I've had for weeks.
I worked for local government for a while and was told "We need to be on Facebook / Twitter" by people who didn't know what Facebook and Twitter were.

I'm sure the same naivety came into play here. Just because you want to engage with people doesn't mean that they are desperate to engage with you. At least not in the way you want.

Of course users were going to find creative ways to abuse the thing. Of course there were going to be bugs and unforeseen / poorly tested trains of events causing awkward error messages.

Read on:
https://www.theguardian.com/technology/2018/feb/01/matt-hancock-mp-app-released

Wednesday, 31 January 2018

May the force not be with you (rumoured dropping of 3D Touch / force touch)

I don't like rumours but this one seems to have some credibility. It has split opinions, with some saying that they're not surprised; it's something you have to learn but can be useful when you have. Others say that it's a make-or-break feature for them. Others (I guess the majority) being unaware of how to use it properly or that it even exists.

I'm not surprised that it hasn't caught on. A long or harder press isn't analogous to 'give me a different action'. If you watch my Mum use her iPad, you'll see that it's analogous to 'shouting' the same thing that you just asked for because you didn't get a result the first time, or didn't get the result you expected. If my Mum received a popup menu or some other action as a result of trying again but harder, she'd find it confusing.

I never press to click my laptop trackpads, I have 'tap to click' switched on and hold the ctrl key if I want a context menu.

I think the 3D touch feature on the handheld devices is an unintuitive move that Micro$oft would be proud of. If Apple are dropping it, it'll be for a good reason. Maybe, like the also-rumoured dropping of OLED screens, it's a cost-benefit decision. (ie more cost but with few users feeling the benefit.)

Further reading
https://mashable.com/2018/01/29/apple-3d-touch-iphone-dead-rumor

Sunday, 28 January 2018

Wot no toilet paper emoticon? (Emoji as a universal language)

Without going into the details of why I wanted a roll of TP emoji, I was amazed to discover that there isn't one - yet.

Back in the day I used to add a ;-) to avoid misunderstandings in emails and texts. But I'm in the "really???!" camp when it comes to adding a string of pictures at the end of a text or social media post.

When Craig gave that infamous demo of the iPhone X ("let's go to backup") He said something along the lines of "Here we are with the most advanced technology in a phone and you want me to make a poop sing" (I can't find the clip and get the exact quote). The technology has well and truly overtaken our real needs.

But then while looking at the page about the forthcoming toilet roll emoji it occurred to me that we're making a great job of assigning images to objects with a short code for each. We're building a dictionary of bytecodes for pictorial representations of objects.

That's incredibly efficient and has potential beyond embellishing your texts. How soon before we start adding more abstract concepts like 'week' and 'meeting', adjectives and conjunctions etc  and end up with a universal language which can be stored incredibly efficiently?

Friday, 19 January 2018

Limiting Integrity / Scrutiny's scan to a particular section of the site or directory

This is a frequently-asked question and a subject that's been on the workbench this week.

Integrity / Integrity Plus and Scrutiny allow you to set up rules which limit the scan, or prevent checking of urls matching a pattern:


The rules dialog* will shortly look as above; the dialog will appear as a sheet attached to its main window, it'll allow 'urls which contain' / 'urls which don't contain' and the 'only follow' option is gone from the first menu, leaving 'ignore', 'do not check', and 'do not follow'.

('only follow' was confusing because if you have two such rules, then it doesn't make logical sense. 'do not follow urls that don't contain' does the same job and makes sense if you have more than one.)

It's important to say that it shouldn't be necessary to use a rule such as 'do not follow urls that don't contain /mac/scrutiny'  if you are starting at peacockmedia.software/mac/scrutiny  because Integrity and Scrutiny should only follow urls at or below the 'directory' that you start at.

It's also important to note that 'not following' isn't the same as 'ignoring'. If a link appears on a page then you will see it in the link check results regardless of whether it matches your 'follow' pattern. If you only want to see urls which follow a certain pattern, use an 'ignore' rule instead.

(That last statement applies to the link check results. The SEO table in Scrutiny and Sitemap table in Integrity Plus and Scrutiny should only show pages which meet your rules and appear at or below the directory you start in.)

Important note - if you're 'ignoring urls not containing'  then you probably won't be able to find broken images or linked files (unless their urls happen to match your rule's pattern). So if you have the 'images and linked files' option checked then you'll need to use a 'follow' rule rather than 'ignore'.

Protip - this has been possible for a while but not very well documented: you can use certain special characters. An asterisk to mean 'any number of any character' and a dollar sign at the end to mean 'appears at the end'. For example,  "don't check urls that contain .dmg$" will only match urls where .dmg appears at the end of the url. And peacockmedia.software/*/mac/scrutiny will match  peacockmedia.software/en/mac/scrutiny  and peacockmedia.software/fr/mac/scrutiny

Regex is not supported (yet) in these rules (though it is available in Scrutiny's site search). It's not necessary to type whole urls or to use the asterisk at the start or end of your term. By default it's a partial match, so "urls not containing /fr/mac/scrutiny" will work.


The new "urls that don't contain" functionality is in Scrutiny from v7.6.8  and should shortly be available in Integrity and Integrity plus from v6.11.14


*Years ago, these rules were called 'Blacklist / Whitelist Rules'. I'm mentioning that in case anyone searches for those terms.

Monday, 8 January 2018

Division by zero

I've just watched this video which was embedded in this Quora answer about division by zero. The video is so entertaining I had to share it. A guy is asked to perform a division by zero on his mechanical calculator. Is it dangerous? Does the machine catch fire?
I wonder whether it occurred to Charles Babbage to try this with his difference engine.

On today's computers I'm surprised that this isn't dealt with at a high or low level. An attempt to divide by zero can crash an app and it's surprisingly easy to do. Should it be necessary for a developer to have their wits about them every time they perform a division using a divisor that is a variable? And have to write an extra line to check that that variable is not set to zero?

It would perhaps be more elegant if the processor returned the largest possible number (INT_MAX or FLT_MAX etc depending on the type of the divisor). Yes, that's the wrong answer, but we're working with approximations all the time - for example,

float f = 1.0;
if (f == 1){
 // unlikely to be executed because f as a float isn't exactly 1 - it has a 'precision',
 // in terms of its bits it'll probably be 0.999999999999 or something
}

There are some very interesting philosophical points in Dave Williamson's answer. Such as the fact that infinity isn't a number. Zero is more generally considered a number, but that is questionable.

All of this inspired me to draw this:
Science Museum, year 2422. Yes, this is Babbage's original difference engine. At the unveiling, he mistakenly gave it a calculation involving a division by zero. Since he hadn't thought to put any kind of cancel button, it's been chugging away pointlessly ever since