New Trains to get Longer Numbers

[PenrithBeacon]

In the early 70's I'm not sure what else it could have been hosted on, and I think that memory wasn't squandered in those days whatever you were running a program on.

 

(E.g. storing years as a two digit number and the "millennium bug")

During the mid-late 20thC nobody thought much about any other century certainly not the next one. The millennium 'bug' wasn't much of a bug at all, unless applications hadn't been migrated to later platforms where the form 'yyyy' was mandatory. I remember I did a lot of millennium testing (all hands to the wheel) and didn't find one snag, not one. Neither did anyone else in the firm I worked for, AFAIK.

[Coryton]

3. Hiding data from users within a database is hardly anything unusual.

 

Agreed. But why stick numbers on the side of a locomotive with a space in but go to extra effort to run the two together in TOPS if they're stored separately? (As you've pointed out, modifying code in those days was a non-trivial exercise).

 

Why integers for the numbers, its conceivable its for memory space reasons, rather than arithmetic, but I assume there is a valid argument for Number, as the D and E prefixes were dropped, Were the region letters stored in TOPS for DMUs etc ?

 

Yes TOPS did store regional prefixes, and the alphabetic prefixes to wagon numbers. But they went into a different database to locomotive numbers.

 

Bear in mind that TOPS also handled the old four digit locomotive numbers (Hymeks and Warships didn't get allocated "TOPS" numbers) so if TOPS did handle class numbers separately it must have done it in a quite sophisticated way.

[adb968008]

I’ve decided to read up a bit more, I recommend the wiki article, which I haven’t hithero read but supports my conclusion...

https://en.m.wikipedia.org/wiki/TOPS

 

“The first two or three digits were used to denote the class of locomotive or multiple unit. The numbers were often written in two space separated groups, such as "47 401" to highlight that division, but the TOPS system actually stored and displayed them without the space: ”

 

Additionally, and as would make sense in such a database normalisation would have its limit..

 

“Sequentiality was all that was required, but with the requirement to renumber, it was decided to adopt a logical system for classification, and the five- or six-digit TOPS number was divided into two parts”

 

Unless the wiki article is incorrect (wiki can be), but my reading and comparing with my own experience of equivalent systems I would tend to concur.

 

The example tops report displayed is a composition screen, the data is not coming from a single table.

It looks to be several differing tables, keyed together and the resulted query dataset is composed in a “readable” format on the screen.

 

It’s ancient, but it still works. Today such a system would have a nice UI with lots of fields, presumably in a browser with an api or a direct connection to the database.

Edited February 5, 2018 by adb968008

[frobisher]

Now computer memory and storage was precious in those days and the easiest way to store numbers in a compact form is to store them as numbers. If TOPS ran on an 8 bit machine, then two bytes would give numbers from 0 to 65535. So maybe it was both a 4 and a 5 digit system - any 4 digit number you like and 5 digits if not too large? BR TOPS would then have required an extra digit giving far more numbers than needed.

 

Then again I don't know how unit numbers are handled - do these 6 digit numbers go in the locomotive file? Or somewhere else? Or not on TOPS at all? Anyone?

Your two byte solution fails out of the gate because of the electrics up in the 80's, and you're thinking in a way too modern way for the issue at hand; Packed Binary Coded Decimal (https://en.wikipedia.org/wiki/Binary-coded_decimal) is probably your bunny here.

[The Stationmaster]

I’ve decided to read up a bit more, I recommend the wiki article, which I haven’t hithero read but supports my conclusion...

https://en.m.wikipedia.org/wiki/TOPS

 

“The first two or three digits were used to denote the class of locomotive or multiple unit. The numbers were often written in two space separated groups, such as "47 401" to highlight that division, but the TOPS system actually stored and displayed them without the space: ”

 

Additionally, and as would make sense in such a database normalisation would have its limit..

 

“Sequentiality was all that was required, but with the requirement to renumber, it was decided to adopt a logical system for classification, and the five- or six-digit TOPS number was divided into two parts”

 

Unless the wiki article is incorrect (wiki can be), but my reading and comparing with my own experience of equivalent systems I would tend to concur.

 

The example tops report displayed is a composition screen, the data is not coming from a single table.

It looks to be several differing tables, keyed together and the resulted query dataset is composed in a “readable” format on the screen.

 

It’s ancient, but it still works. Today such a system would have a nice UI with lots of fields, presumably in a browser with an api or a direct connection to the database.

 

As the 'logical system for classification' predated TOPS by 5 years (and had been in everyday use for operational purposes since 1968) at least part of what is written in that article is incorrect.

[Coryton]

Your two byte solution fails out of the gate because of the electrics up in the 80's, and you're thinking in a way too modern way for the issue at hand; Packed Binary Coded Decimal (https://en.wikipedia.org/wiki/Binary-coded_decimal) is probably your bunny here.

 

Ah yes. I dismissed packed solutions as being too computationally intensive, but I'd forgotten that BCD support would likely have been built-in. Not much call for it these days.

 

Seems far more plausible than using integers.

 

 

As the 'logical system for classification' predated TOPS by 5 years (and had been in everyday use for operational purposes since 1968) at least part of what is written in that article is incorrect.

 

I recall one of the wikipedia pages on TOPs making the "numbers started at 001 because TOPS didn't like zeros" claim but I can't find it now. It may have been removed. Or I may have missed it. Or mis-remembered.

 

Searching for information on TOPS is quite hard because it's a rather common word...

Edited February 6, 2018 by Coryton

[iands]

Ah yes. I dismissed packed solutions as being too computationally intensive, but I'd forgotten that BCD support would likely have been built-in. Not much call for it these days.

 

Seems far more plausible than using integers.

 

 

 

 

I recall one of the wikipedia pages on TOPs making the "numbers started at 001 because TOPS didn't like zeros" claim but I can't find it now. It may have been removed. Or I may have missed it. Or mis-remembered.

 

Searching for information on TOPS is quite hard because it's a rather common word...

Hi Coryton,

 

I'm not sure that TOPS didn't/doesn't like zeros. Reading a copy that I have of "TOPS The Story of a British Railways Project" by Robert Arnott, the TOPS Project Manager (Chief Operations Manager, BRB). He makes no mention of any difficulties within TOPS of numbering locos 'xx000'. By the same token, there is no explanation as to why locos started 'xx001' either. So I imagine this might just have been a "Project" decision rather than a software handling issue.

 

It is interesting to note that the numbering of 5 digit TRAs (TOPS Responsibility Areas) is comprised as follows:

 

First two digits = TRA or group of TRAs in the same working complex,

Third digit = section of line within that TRA or group,

Last two digits = individual locations on each section of line.

 

The individual location number range is from '00' to '99'. So if there was/is no problem within TOPS to having a location code as '00' then I can't see a problem with a loco having a number '000'. Incidentally, within the document any reference to a loco number is a 5 digit number with no spaces, e.g. 55001, and not 55 001.

 

Hope this helps in some way.

 

PS. I find the wiki TOPS explanations somewhat brief to say the least, however, this may be down to the fact that the "author" was having some difficulty in sourcing relevant information. What I find slightly misleading in the article is that it implies that SP had TOPS from the early 60s. Not quite, yes they had the idea in the very early 60s, and a few years in development but their first system didn't "go live" until 1968.

 

Regards, Ian.

[Coryton]

Hi Coryton,

 

I'm not sure that TOPS didn't/doesn't like zeros. Reading a copy that I have of "TOPS The Story of a British Railways Project" by Robert Arnott, the TOPS Project Manager (Chief Operations Manager, BRB). He makes no mention of any difficulties within TOPS of numbering locos 'xx000'. By the same token, there is no explanation as to why locos started 'xx001' either. So I imagine this might just have been a "Project" decision rather than a software handling issue.

 

It is interesting to note that the numbering of 5 digit TRAs (TOPS Responsibility Areas) is comprised as follows:

 

First two digits = TRA or group of TRAs in the same working complex,

Third digit = section of line within that TRA or group,

Last two digits = individual locations on each section of line.

 

The individual location number range is from '00' to '99'. So if there was/is no problem within TOPS to having a location code as '00' then I can't see a problem with a loco having a number '000'. Incidentally, within the document any reference to a loco number is a 5 digit number with no spaces, e.g. 55001, and not 55 001.

 

Hope this helps in some way.

 

PS. I find the wiki TOPS explanations somewhat brief to say the least, however, this may be down to the fact that the "author" was having some difficulty in sourcing relevant information. What I find slightly misleading in the article is that it implies that SP had TOPS from the early 60s. Not quite, yes they had the idea in the very early 60s, and a few years in development but their first system didn't "go live" until 1968.

 

Regards, Ian.

 

Thanks.

 

I don't think TOPS does hate zeros...I meant that as an example of why I wasn't convinced that the Wikipedia site was correct.

 

If the page is in error, you can of course correct it (without even the effort of making an account - just go for it!)

[The Stationmaster]

Hi Coryton,

 

I'm not sure that TOPS didn't/doesn't like zeros. Reading a copy that I have of "TOPS The Story of a British Railways Project" by Robert Arnott, the TOPS Project Manager (Chief Operations Manager, BRB). He makes no mention of any difficulties within TOPS of numbering locos 'xx000'. By the same token, there is no explanation as to why locos started 'xx001' either. So I imagine this might just have been a "Project" decision rather than a software handling issue.

 

(snipped)

Regards, Ian.

 

I suspect it is far more likely to have been an M&EE decision rather than anything else.  Don't overlook the fact they had been hankering after full renumbering around their class identifier system some years before TOPS was even being thought of and if they were thinking about it it seems not unlikely (but I have no evidence to support this view) that they had also given consideration to how number series would work within class identifiers and possibly even how sub-series would be identified.  

 

Quite obviously - whatever took place - any system of renumbering would have deeply involved the M&EE people and quite probably some input from operators as well although as far as the operating side was concerned numerous people had been quite happily working with the class identifier number (only) since 1968.

[Coryton]

I suspect it is far more likely to have been an M&EE decision rather than anything else.  Don't overlook the fact they had been hankering after full renumbering around their class identifier system some years before TOPS was even being thought of and if they were thinking about it it seems not unlikely (but I have no evidence to support this view) that they had also given consideration to how number series would work within class identifiers and possibly even how sub-series would be identified.  

 

Quite obviously - whatever took place - any system of renumbering would have deeply involved the M&EE people and quite probably some input from operators as well although as far as the operating side was concerned numerous people had been quite happily working with the class identifier number (only) since 1968.

 

One thing that's a little odd about "TOPS" numbers is the leading zero in class numbers, e.g. class 03 and 08 not 3 and 8 which might seem more natural.

 

I presume this is down to the first digit describing the power category though, and again not something resulting from a computer system which insisted on 5 digits. I wonder though...did 4 digit numbers for Westerns need a leading 0 to make TOPS happy? If not, it would have needed to cope with spaces as well as numbers in the locomotive file. But I digress.

 

The power category is an example of starting to count at zero, but perhaps with precedent here - didn't the LMS have an 0F power category?

[The Border Reiver]

One thing that's a little odd about "TOPS" numbers is the leading zero in class numbers, e.g. class 03 and 08 not 3 and 8 which might seem more natural.

 

I presume this is down to the first digit describing the power category though, and again not something resulting from a computer system which insisted on 5 digits. I wonder though...did 4 digit numbers for Westerns need a leading 0 to make TOPS happy? If not, it would have needed to cope with spaces as well as numbers in the locomotive file. But I digress.

 

The power category is an example of starting to count at zero, but perhaps with precedent here - didn't the LMS have an 0F power category?

 

Database sorting order. If it was just say 3 you would get 24, 25, 26, 27,29, 3, 31...

[adb968008]

Database sorting order. If it was just say 3 you would get 24, 25, 26, 27,29, 3, 31...

 

Only if the datatype was a character.

If it was sorted as a number, or was a numeric, it would not.

 

It is an interesting thought how westerns were stored within TOPS, whether they were allocated 52 xxx or not. I'd guess they were and common sense was applied at the depot, and what number was D1000 ? - though BR Database suggests it was withdrawn just before TOPS.

 

I find it fascinating that people think the decision to start at 1 as an arbitrary uninformed finger in the wind project decision to deliberately cause confusion on the various sheds around the country with numbers like (D200)40122 and (D5500) 31018, (D5519) 31101 [not 31100], 20050 (D8000), 37350 (D6700), 47401 (D1500) [Not 47400] etc etc.. A lot of people would be paid good money for such decisions, and there would be several meetings to discuss various options and competing theories to reach it.. its refreshing to think all that would be swept aside on a project which required government approval just by tossing a coin and saying heads (0) or tails (1)...

 

There was a reason for this as a standard, At a minimum it ensures consistency, I'm sticking to my belief it's IT related, not just experience but common sense.. you wouldn't upend the loco fleet nationwide without good reason. Most likely the reasons will still be documented in the archives given the cost / profile / government involvement in this project. 

 

Ultimately there is no right or wrong answer, just a decision to create a standard. The discussion about 0 or 1 as a measure starting point will rumble on for eternity unless a standard is agreed and documented. Its the bane of IT for 50 years and its such a simple starting point. Several times I've been able to kill a meeting for a few hours by starting on that very subject.

Edited February 7, 2018 by adb968008

[Coryton]

Database sorting order. If it was just say 3 you would get 24, 25, 26, 27,29, 3, 31...

 

There was a reason for this as a standard, At a minimum it ensures consistency, I'm sticking to my belief it's IT related, not just experience but common sense.. 

 

Possibly....but as the class numbers were defined some years before locomotive numbers were computerised (and indeed before it was common for anything to be computerised) that would suggest an impressive bit of foresight.

 

(Of course maybe the numbers were originally like that and changed from 3 to 03 etc. in a very minor change to keep TOPS happy).

 

, not just experience but common sense.. you wouldn't upend the loco fleet nationwide without good reason.

 

But the good reason might have been that people wanted a better numbering system, not that they went out and bought a computer system which then forced them to renumber everything. (I believe that TOPS was extensively reprogrammed in the UK - I would venture to suggest that if there was an incompatibility, it might have been cheaper to reprogram TOPS than choose a new number for every locomotive and stick it on the side).

 

 

It is an interesting thought how westerns were stored within TOPS, whether they were allocated 52 xxx or not. I'd guess they were and common sense was applied at the depot, and what number was D1000 ? - though BR Database suggests it was withdrawn just before TOPS.

 

From what I've read, they just went into TOPS as is, along with Hymeks. It doesn't make sense that they would all have two numbers - one on the nice cast iron plates and another to type into TOPS that had to be looked up every time because it wasn't on the side of the locomotive.

 

Were 1001 and 1002 also withdrawn in time? Either TOPS could distinguish between 01001 and 1001, or there would be a clash otherwise.

[The Stationmaster]

One thing that's a little odd about "TOPS" numbers is the leading zero in class numbers, e.g. class 03 and 08 not 3 and 8 which might seem more natural.

 

I presume this is down to the first digit describing the power category though, and again not something resulting from a computer system which insisted on 5 digits. I wonder though...did 4 digit numbers for Westerns need a leading 0 to make TOPS happy? If not, it would have needed to cope with spaces as well as numbers in the locomotive file. But I digress.

 

The power category is an example of starting to count at zero, but perhaps with precedent here - didn't the LMS have an 0F power category?

 

The Eastern Region diesel loco class numbering system worked very much on a power rating basis although it had difficulty dealing with a multitude of classes and variations within them and asa. result got a bit complicated.  The BR 1960s system appears to have been based on a similar basic idea although there are obvious exceptions such as a Type 1 numbered in the 2X series (Class 20) and a Type 4 (D10XX) allocated to the 5X series - basically i presume because there were insufficient digits in the second position to cater for know types and any anticipated developments.

 

And don't forget that leading zero predated TOPS by several years - never forget the class identifier system was there well before TOPS was even being seriously considered by BR.

 

In my experience with TOPS (but before renumbering got underway although later power reports still worked in the same way) is that it accepted D10XX numbers as 4 digit items although maybe it automatically did something to them within the system that wasn't visible to the user - to answer that we need someone who was involved in programming the system, er 40+ years ago.

 

 

Only if the datatype was a character.

If it was sorted as a number, or was a numeric, it would not.

 

It is an interesting thought how westerns were stored within TOPS, whether they were allocated 52 xxx or not. I'd guess they were and common sense was applied at the depot, and what number was D1000 ? - though BR Database suggests it was withdrawn just before TOPS.

 

I find it fascinating that people think the decision to start at 1 as an arbitrary uninformed finger in the wind project decision to deliberately cause confusion on the various sheds around the country with numbers like (D200)40122 and (D5500) 31018, (D5519) 31101 [not 31100], 20050 (D8000), 37350 (D6700), 47401 (D1500) [Not 47400] etc etc.. A lot of people would be paid good money for such decisions, and there would be several meetings to discuss various options and competing theories to reach it.. its refreshing to think all that would be swept aside on a project which required government approval just by tossing a coin and saying heads (0) or tails (1)...

 

There was a reason for this as a standard, At a minimum it ensures consistency, I'm sticking to my belief it's IT related, not just experience but common sense.. you wouldn't upend the loco fleet nationwide without good reason. Most likely the reasons will still be documented in the archives given the cost / profile / government involvement in this project. 

 

Ultimately there is no right or wrong answer, just a decision to create a standard. The discussion about 0 or 1 as a measure starting point will rumble on for eternity unless a standard is agreed and documented. Its the bane of IT for 50 years and its such a simple starting point. Several times I've been able to kill a meeting for a few hours by starting on that very subject.

 

Nothing about numbers would cause confusion to anybody other than those actually applying the new numbers or trainspotters.  Railwaymen out on the ground work on the basis of the painted number - end of story (although of course in the steam age the prefix added to former Big Four numbers was very often permitted on their home Region or at sheds in that area).  But with TOPS it didn't matter - the number that was input to the system for whatever reason was the one on the side of the loco and I doubt if anyone gave two hoots about what number that loco had carried the week before unless it was to carry forward the reputation of a bad 'un.

 

Basically somebody somewhere decided, logically in my view, not to have locos numbered zero - end of story.

[Coryton]

And don't forget that leading zero predated TOPS by several years - never forget the class identifier system was there well before TOPS was even being seriously considered by BR.

 

Yes you've said that the class identifier system pre-dates TOPS.

 

But - in principle - it might have been that shunters were originally referred to as class 1, 3, 8 etc. in the class identifier system and to keep TOPS happy the numbers had to be very slightly changed to start 01, 03, 08 etc. 

[The Border Reiver]

Having spent 20 years working with SQL databases the rule of thumb is that if it's not something you will ever need to do math on, you don't store it as a number. Thus to get the correct collation for a varchar field, one would need to use leading zeros.

[HillsideDepot]

I don't recall where I originally saw this TOPS printout, so I unfortunately cannot credit the photographer. However, it shows 1013 on the famous day that it worked to Leeds and back.

There it is, slightly below half way down.

 

I've no idea what clever stuff might have gone on inside TOPS to show it, but it is reported as a four figure number.

[adb968008]

I don't recall where I originally saw this TOPS printout, so I unfortunately cannot credit the photographer. However, it shows 1013 on the famous day that it worked to Leeds and back.

D1013 at Leeds.jpg

There it is, slightly below half way down.

 

I've no idea what clever stuff might have gone on inside TOPS to show it, but it is reported as a four figure number.

Very nice.

So it wasn’t as a class 52, at least in the fleet number field.

I’ll bow to facts on the 52.

 

So we learn a little more.. The was most likely 3 bytes used in the fleet number field.

Pity there no data field in that report.

Edited February 7, 2018 by adb968008

[iands]

I don't recall where I originally saw this TOPS printout, so I unfortunately cannot credit the photographer. However, it shows 1013 on the famous day that it worked to Leeds and back.

D1013 at Leeds.jpg

There it is, slightly below half way down.

 

I've no idea what clever stuff might have gone on inside TOPS to show it, but it is reported as a four figure number.

 

Perhaps the "clever stuff" was similar to resolving the problem of wagons between Dover and Dunkerque by putting the boats that carried the ferry wagons on the locomotive file by using the physical characteristics of the locomotives which had been scrapped. This was achieved by using the 'Diagram Number' field with the name of the ferry; for example 99003 'dummy' locomotive number was applied to the "Essex Ferry". In response to a locomotive enquiry for that number the abbreviation ESSEXFRY would be shown under the 'Diagram Number'. So an enquiry which asks the computer to provide details of locomotives approaching a TRA made at Dover would include boats as well as locomotives.

 

The above is a slightly edited version from the "TOPS Story" I mention in #208 above. I imagine that if they could devise a method of displaying a ferry in the database, it wouldn't be a huge task to devising a similar method to display "1013" as per the example in #218 by utilising other 'dummy' locomotive numbers.

 

Regards, Ian.

[Coryton]

I don't recall where I originally saw this TOPS printout, so I unfortunately cannot credit the photographer. However, it shows 1013 on the famous day that it worked to Leeds and back.

D1013 at Leeds.jpg

There it is, slightly below half way down.

 

I've no idea what clever stuff might have gone on inside TOPS to show it, but it is reported as a four figure number.

 

Fascinating.

 

So it seems that TOPS could (and no doubt still can) distinguish between 0 and space, at least in the first column.

 

So maybe not BCD, and maybe it could indeed cope with alphanumerics.

 

In that case even the electric locomotives could have gone straight onto TOPS as-is, with the exception of the 76's with their longer 1948 numbers which by then seem to have acquired an E at the start - but had the "TOPS" numbering not come in wouldn't have been unreasonable to renumber these as they were in a previous scheme.

[frobisher]

One thing that's a little odd about "TOPS" numbers is the leading zero in class numbers, e.g. class 03 and 08 not 3 and 8 which might seem more natural.

 

I presume this is down to the first digit describing the power category though, and again not something resulting from a computer system which insisted on 5 digits. I wonder though...did 4 digit numbers for Westerns need a leading 0 to make TOPS happy? If not, it would have needed to cope with spaces as well as numbers in the locomotive file. But I digress.

I suspect it's even easier to explain than that, especially given the printout linked. If both the characters of the "class" portion of the number (assuming XX YYY) are numbers then it's a proper TOPS class number in the new scheme, otherwise it's a legacy number (with presumably some look up table shenanigans needed to identify the class of the loco).

 

Were the Peaks renumbered in full ahead of TOPS taking on the locos? I can see that less than 3 digits in the YYY part causing "issues".

[frobisher]

In that case even the electric locomotives could have gone straight onto TOPS as-is, with the exception of the 76's with their longer 1948 numbers which by then seem to have acquired an E at the start - but had the "TOPS" numbering not come in wouldn't have been unreasonable to renumber these as they were in a previous scheme.

I seem to recall the 76's got a quick and dirty renumbering with 7's merely being stickered over the leading E2's.

[frobisher]

Yes you've said that the class identifier system pre-dates TOPS.

 

But - in principle - it might have been that shunters were originally referred to as class 1, 3, 8 etc. in the class identifier system and to keep TOPS happy the numbers had to be very slightly changed to start 01, 03, 08 etc.

I'm pretty sure the leading 0 was always part of the numbering scheme - I'm sure someone shared a document from the 60's in respect of showing the 33/1 from 34 change, and the shunters had their leading zeroes. It makes sense to differentiate the number scheme from the previous regime (if it hadn't been for those pesky Tommies...).

[caradoc]

So, did BR actually need to renumber its loco fleet at all when TOPS was introduced ? If they did not it would have saved me an awful lot of hassle (I couldn't afford then to buy a railway magazine each month, so I used to cycle to the Oxford Library at Westgate and copy down the list of renumbered locos from Modern Railways, which was available in the Reading Room !)

[adb968008]

So, did BR actually need to renumber its loco fleet at all when TOPS was introduced ? If they did not it would have saved me an awful lot of hassle (I couldn't afford then to buy a railway magazine each month, so I used to cycle to the Oxford Library at Westgate and copy down the list of renumbered locos from Modern Railways, which was available in the Reading Room !)

That TOPS report did leave a clue.. but a contradiction too...

 

1013 was minus the ‘D’. Which suggest the system is integer based, and so letters have to go.

This could be backed up by “E” numbers on electrics, which if it was a char based data set, could have been retained and distinguish diesel from electric... as “E” numbers went.. there is duplication, so I guess they system was integer based and so electrics were renumbered to avoid it...

 

But...

 

An integer “03” for an “03”.. is just “3”.. the zero is normally dropped.

This means it could be a char based dataset, or special coding to insert the leading zero...

 

Back to stage 1.