The Hippo Techo Blog

Close ...
Mac Mini 6,2 Late 2012 model running OSX 10.9.1 (new not an upgrade)

Retired the ol’ P4 (Brilliance 17” monitor from 1996 still going strong!) and put a wee Mini on the shelf.
Mini looks good and very upgradeable which is more than can be said for most of Apples newer offerings aside from the Mac Pro. Good amount of ports, most of which are in use except Firewire.

Mavericks:::
Not too bad, the scrolling direction is reversed but is still optional and basically seems to work with older stuff in the network OK. Very happy to see that YWE/Konfabulator widgets still work and the LG DVD-RAM drive appears to function OK.

The main issue I’ve found (could be a pun there!) is the dodgy handling of external storage drives.

Issue 1::
I moved a WD Mirror Edition RAID 1 drive from the old P4 onto the Mini. The drive was formatted as NTFS and could be mounted/read by OSX no problem but I decided to re-format it as Journaled HFS+ using the WD Drive Manager app (default name – MyBook). This was no problem and worked fine but I did find that intermittently the drive would be dismounted and a ‘Incorrect Disk Eject’ error was produced. This does not inspire confidence in the storage system.
The forums are ablaze with reports of lost data on WD drives after upgrading to Mavericks so there is a problem there somewhere.
I had a play round with the Energy Saver settings in the System Preferences and have had some limited success.

Fix: Unchecked ‘Put hard disks to sleep when possible’ setting – this appears to have stopped the random dismounting issue.

Issue 2::
Lacie 2big RAID would sporadically start c. every 70 sec or so and remain on for 1 min

Fix: Unchecked ‘Wake for network access’ in Energy Saver panel.

Issue 3::
LG external drive (LG BE08LU20) sporadically turns on/off (no disk in drive) when Mini in sleep mode.
Not sure what was happening here but when a disk is inserted the drive became very active for a long while – it looks like Spotlight was the culprit as it was indexing the disc. Messages relating to the .`Spotlight-V100’ store were filling the Console.

Fix: Disabled Spotlight indexing of the DVD volume.

How: Use Terminal
Enter - sudo mdutil –i off “/Volumes/MEI_UDF”

Get - Password: (need admin level password here)
Then get - /Volumes/MEI_UDF:
Indexing disabled (took a while since I did this while it was indexing)

Issue 4::
Using Finder quite often the contents of folder would be very slow to show – makes the system feel like a dog!

Fix: Again in the forums there seems to be reports of this as being an issue. It seems to have improved since doing an SMC reset on the Mini (basically switch off and unplug for 30secs then replug and wait 5secs then start up again).


With hindsight I would probably forgo the Fusion Drive for a large ordinary drive since the fast part of the Fusion Drive is quickly used and so a lot of data transfer will happen between the SSD and HDD. I think it would be better if the user could specify what type of data should reside on the fast part of the volume and in the future when larger/cheaper SSD’s are available just replace the Fusion Drive.

Update:

Still getting the clicking on/off with the ROM drive.
At one point I inserted a CD into the drive whereupon the WD RAID dismounted with a message saying ‘Not ejected properly’. I checked the LaCie Raid Monitor and it said ‘No compatible RAID detected’ although the RAID was still functioning OK.
A restart didn’t bring back the WD RAID but a repower did.

The solution appears to be to uninstall all the manufacturer monitor apps which is inconvenient but they do seem to be incompatible with Mavericks.

Update to 10.9.2 went OK

Close ...
OSX 10.6.8 Snow Leopard.
Had a problem with a crash when a New Tab was opened by clicking the New Tab stub or via the Menu. Tabs would open OK if done from a link on another page. Problem still existed in Safe Mode with all Extensions/Plugins disabled
(implemented via Terminal - /Applications/Nightly.app/Contents/MacOS/firefox -safe-mode ).
Backed up the bookmarks/etc and did a Reset Firefox that was offered from the crash restart screens. Reset worked great and the Nightly came back with the Plugins and bookmarks intact and the Extensions cleaned out. Just needed to re-download the extensions and everything was good.

For general info on OSX –

Profile folder located at - /Library/Application Support/Firefox/Profiles
Extensions located in above folder under something like - dgte7tfw.default-1387301888020/extensions
Bookmarks these days are held in a sqlite file but can be exported as HTML via the Show All Bookmarks which gives the Library. The Library has a wee button which provides options for –
Backup and Restore, Import/Export as HTML and Import from another browser.

Looks like the Reset option is a good way of cleaning out an old (maybe archaic) installation without losing all history etc.

Close ...
What a saga this can be – seems to very difficult to determine the filter arrangements for Leica lenses – the technical manuals are not much help and the filter manufacturers seem to be accomplices in the darkness.

However just in case anyone needs to know –

X Vario lens filter diameter is 43mm (actually marked inside the lens cap).
The thread pitch may be an issue with certain filters, I could find no reference to the thread pitch but it seems to be 0.75 which I believe is fairly standard these days. I tried a B+W filter with 0.5 pitch and it will go on halfway only and is usable but definitely shouldn’t be tightened further – the correct filter turned out to be a B+W 43ES type with 0.75 pitch, it goes fully on and smoothly as expected.

B+W 43 X 0.5 F Pro MRC 010M UV-Haze filter Part No. 40228 - not correct for X Vario, does not impede zoom.

B+W 43ES 0.75 MRC 010M UV-Haze filter Part No. 23185 – correct for X Vario, does not impede zoom.

Zeiss T* 43 – I believe these work also so must be 0.75 pitch.

Some reviews I read on the X Vario suggested that there was no possibility of fitting a filter – I can only assume that maybe some early prototype versions weren’t threaded.

Update:

B+W 43mm XS-Pro Digital Clear MRC Nano Filter (007M) Item Number 45384 works fine and is half as thick as the regular filters.

Close ...
Recently been having issues with saving web pages – if ‘Save Web Page As Complete’ is done the page is not saved and the save download does not complete.

The problem turned out to be with the Mozilla Archive Format extension (v 3.0.0). This extension allows saving in .maff or .mhtml form but with recent versions of FF it seemingly breaks the Save Complete option.
Isolating the problem involved starting FF in Safe Mode which disables all the extensions and after finding that the Save then worked I had to go through the extensions disabling/enabling until the baddie was found.

Once the MAFF extension was found to be the baddie I then looked at the Preferences of the extension.
The fix proved to be the ‘When Saving A Complete Page’ option – the default was set to ‘Take An Exact Snapshot’ so I set it to ‘Use Browsers Standard Save System’ and it all worked fine.

Close ...


Konfabulator has gone down the tubes (aka. Smart TV using Yahoo Widgets for TV).

This is old news since development of the Yahoo Widget Engine in its desktop form ceased a while back but the runtime engine (version 4.5.2) is still available on the net although the Widget Gallery was closed as well.

Recently it seems the last nail has gone in with the old Konfabulator User Forums now gone.

I find it a real shame that YWE (Konfabulator) has gone simply because it is a wonderful development tool for making web/desktop apps that will work on both Windows and OSX and only really required javascript knowledge without having to get heavily involved in the usual developer languages.

A big advantage of YWE is that the apps work consistently across the platforms unless a really platform specific feature is attempted and the docs are excellent – a real shame that great programs get taken over by big companies that just ruin them

Currently my widgets run on Win XP and OSX 10.6.8 (Snow Leopard) but not sure about future releases. I would like to think that YWE may still work on 10.9 (Mavericks) but we will see – on OSX it seems that it may work OK with 10.7 (Lion) but have issues with 10.8 (Mountain Lion). On Win I believe the YWE still works on Win 7 and Win 8.

I am currently trying Q7Basic (Basic for QT) as a replacement for the YWE tool but having mixed success. It is quite good for doing my BASIC apps and is cross-platform but has a few annoying/showstopping bugs which will not be fixed anytime soon since the single developer is a busy man and I don’t have the expertise to get into the Qt C++ which underpins the BASIC environment. Having said that I have ported a quite complicated BASIC app from VB2005/2008 and the Q7Basic environment is a lot less complicated and easy to produce compiled executables.

Link to enthusiast still using YWE - http://lightquick.co.uk/blog.html

UPDATE:
OSX 10.9.1 Mavericks
YWE 4.5 installs OK and widgets appear to run fine. My widgets basically fired up OK with some just needing Preferences set in the widgets.
The BTHomeHub monitor required the wireless status code to be disabled – there seems to be an issue with the 80211 PrivateFrameworks data that I will need to resolve at some stage (it works fine in Terminal).
The WorldRadio has an issue with .asx streams at the moment but the updated Flip4Mac appears OK – again at some stage I will investigate.
So it largely looks like YWE is still a valid dev tool on OSX 10.6 through 10.9

Running on OSX 10.9.1 - Mavericks.



UPDATE:
OSX 10.10.2 Yosemite installed over Mavericks
YWE 4.5 works OK and widgets appear to function.

UPDATE:
Win 10 in VM P11 on Mavericks host
YWE 4.5 works OK and widgets appear to function.
Also OK on OSX 10.11 El Capitan.

Close ...
Tried to install a later version of Wireshark and although it installed OK there was a problem with the WinPcap install so of course there were no interfaces usable in Wireshark. I tried a standalone WinPcap install and it also failed – the install detects existing WinPcap components even though it had previously been uninstalled and was no longer in the Add/Remove programs, etc.

As it happens if the following files exist they need to be removed and then the computer restarted to clean out.

wpcap.dll /system32 CACE driver
npf.sys /system32/drivers Riverbed
Packet.dll /system32 CACE driver

I can only think that the uninstall with earlier versions was not good since I tried a install/re-install with 4.1.3 and the files were nuked OK.

Close ...
Just spent a lot of time running in circles trying to get my head around this.
Basically I was upgrading a bit of code which used MD5 hashing and clear text saved to cookies – time to improve.

Using crypt() as the hashing command this is my experience –

Create Hash -
$hashed_password = crypt($loginpassword, $salt); - this is the fundamental hash command

Verify Hash –
$new_hash = crypt($loginpassword,$hashed_password);
if ($new_hash == $hashed_password) {
// SUCCESS
} else {
// FAIL
}

Or

if (crypt($loginpassword,$hashed_password) == $hashed_password)) {
// SUCCESS
} else {
// FAIL
}


The hashed password can be stored to a cookie/file/db or whatever but may need escaping for certain chars in some cases. The salt including the password hash data is pre-pended to the hash result.
The complete original stored hash can be used for verification – no need to parse out the salt, crypt will do that!
Verification will not give the original password as a result – it verifies the hashes are correct so therefore the password you have presented is correct. Decryption is not an option.
The hash for a particular password will only be the same if the same salt is used.
Usual rules apply regarding quality of passwords.
MD5 is vulnerable these days so I tried SHA256/512 and Blowfish settling eventually for Blowfish hashing.

Typical hash results –

Using mt_rand() to generate salt then hashing with Blowfish

$2y$12$QHVCHtgiJHLPW8bZkL.q/.i57xslCy3ikqcVWOukQebuvf5m0YNuq
$2y$12$QHVCHtgiJHLPW8bZkL.q/B

Cost is 12 (range 04-31) and salt is 22 chrs.

Using mt_rand() to generate salt then hashing with SHA256

$5$rounds=5000$YWCY.pqQUa/0TKyD$4QD95fB87uj9i7Zfc3uCkaV.XTByRJP4z lmjc5lmi01
$5$rounds=5000$YWCY.pqQUa/0TKyD$ - Pre-Hash salt

Cost is 5000 (range 1000 – 999,999,999 ) and salt is 16 chrs.


I spent a lot of time trying to get it all to work and was mislead by a red herring with the Blowfish method.
If you notice the returned hash contains a truncated salt (1 chr) which I thought was a problem – it isn’t because Blowfish does some bit dumping and stuff and so it looks bad. However with the correct login password presented and the full original hash given to crypt the hash should verify OK. The problem I had was that I was inadvertently presenting a hashed login rather than clear text.
The SHA results do not have the truncation issue at all.










Notes:

$5$ - SHA256
$6$ - SHA512
$2a$ - Blowfish early – security hole
$2x$ - Blowfish security patch, backward compatible
$2y$ - Blowfish from PHP5.3.7 onwards (preferable)

A random salt is used so each login password will have a unique hash that is verifiable without having to store the original password.
The salt is stored and is public but is unique for each password when generated.
In terms of attacks both SHA and Blowfish have cost parameters which essentially alter the time taken to create the hash and in a web scenario another device to block attacks is to limit login attempts for a given password for a set time.

Some folk suggest that mt_rand() is not up to the job these days and something like openssl_random_pseudo_bytes() is preferable – cryptographically I would agree but for me mt_rand() is sufficient given that it is only used to prepare the salt prior to hashing – random enough.

There is also some argument about whether it is good to have the salt included in the hash – my feeling is if simple passwords are used then it could be a liability since given the original hash salt along with the correct password and the known hash method verification is easy. I guess the answer is to use a difficult password or separate the salt from the hash.
Another precaution may be to strip the hashing method part of the header for storage and reinstate it on retrieval for verification.

See HERE for a wee test file.

So all in all good hashing can be done simply and for more compatibility there are some good library codes available.

www.openwall.com/phpass/

crackstation.net/hashing-security.htm

Close ...
Just been looking at using SQLite3 as a backend to some web apps and found it all a bit interesting.
It seems that PDO (PHP Data Objects) is a handy interface for numerous types of database but I did find info relating to using SQLite a bit thin in places. Anyway after a bit of trawling and trying I managed to get a handle on basic usage which is not too difficult.

HERE is a wee PHP file (as txt) that encapsulates my trials and may be of assistance to some folk.

Close ...
Just had an adventure pulling apart a dishwasher to find the source of an unusual buzzing sound. The buzzing seemed to occur when the water was being pumped in after the initial pump out of the stale water at the beginning of the cycle.

After cleaning various filters and bits the cause of the problem appeared to be the wee filter on the water inlet valve.

The mesh was covered with a brown residue blocking most of the holes but was easily cleaned – the buzzing ceased so at least I didn’t have to replace the inlet valve or worse still the dishwasher!

Close ...
Just my experience and Parallels seems to suggest that results may be unpredictable if Parallels Desktop for Mac is run at the same time as VirtualBox.
OSX 10.6.8 Snow Leopard on MB Pro

Both guest OS’s are running XP Pro SP3 32bit and both use native disk format (VDI for VB and HDD for Parallels)

Parallels 7 and VirtualBox 4.2.10
Seemed to co-exist quite well and could transfer files OK between each guest using Shared Folders within the guests.
Only glitch was that HTTP browsing from the VirtualBox guest stopped working when Parallels guest in use but seemed to work OK if the VirtualBox guest was started before the Parallels one.

Parallels 8 and VirtualBox 4.2.10
Co-existence not as good – noticed that if Parallels guest was in use that trying to start the VirtualBox guest gave an error – “You can’t open the application VirtualBoxVM because it may be damaged or incomplete”.
I did find however that if the VirtualBox guest is started and in use before the Parallels guest is opened then both guests appear to co-exist OK.