michalfabik's Diary

Recent diary entries

The Proposal

In late 2020, I drafted a proposal to introduce a common format for the name=* tag in the Republic of Srpska, a territorial subdivision of Bosnia and Herzegovina (BiH). The reasons were mainly general data inconsistency and recurring edit wars. There are two official scripts in the Republika Srpska, Latin and Cyrillic, both of which are in active use. Moreover, some mappers (as well as members of the general public) consider the usage of either or both of the two scripts a political tool and/or an instrument of nationalist propaganda. This means that it is hard to reach a consensus as to what the name=* format should be. It can be argued that a tag like name=Бања Лука is violating the On-the-Ground Rule because both scripts - not just Cyrillic - are official. It can also be argued that a tag such as name=Banja Luka / Бања Лука violates the OTG Rule because the city is not called “Banja Luka Slash Banja Luka” or because one of the scripts is “more official” than the other and should therefore be listed first. Therefore, I included several possible name formats in my proposal and invited the members of the OSM community in BiH to vote for the most suitable one with the intention to propose the winning format “officially” to the worldwide OSM community. The voting took place on the proposal’s discussion page.

The Response

I have contacted 55 mappers in total, most of whom were my saved private message contacts with whom I have corresponded over more than 4 years of consistently reaching out to BiH users, others were top-ranking mappers recently active in the country. As you can see on the proposal discussion/voting page linked above, the response was rather underwhelming, with many of the most active users ignoring the communication and the vote completely.

I have, however, received a number of messages whose authors, while stopping just short of being openly offensive to me personally, expressed strong opposition to one or more of the proposed name formats, on political grounds. They asserted that OSM users of their respective “enemy” ethnicities (to whom they did not hesitate to refer with ethnic slurs) were going to “brigade” the vote to make sure that such an “unacceptable” name format wins. In reality, the voter turnout was quite low and nothing of the sort took place.

Interestingly, one of these users mentioned that a similar situation already occurred several years ago and that in the end, the OSM community came to understand that they would incur a lot of trouble if they were to introduce a tagging policy this user perceives as undesirable, effectively saying that the OSM community was bullied into submission by BiH users with a political/nationalist agenda. I couldn’t find anything to corroborate this and I would be most grateful if the Data Working Group (or anybody else involved) could shed some light here, assuming, of course, that anything of that sort actually happened.

What Now

I’m somewhat unsure about hot to proceed at this point. On one hand, the voter turnout was low and I do realize that I wasn’t exactly rigorous in advertising the vote. (None of the communication channels established within the BiH community is frequented by more than some 10 people regularly.) It is to be expected that some will view any action based upon this vote alone as an unwarranted initiative of a few lone busybodies, if not as an outright hostile act. On the other hand, I have been in contact with mappers who agreed that something should be done about the current situation but I have never witnessed any actual attempt at bringing about a solution. Other users still have expressed indifference or explicit refusal to deal with the subject.

I’m writing this as a kind of a backstory to refer the folks on the tagging list to, if/when I officially submit the proposal for OSM-wide voting. Come to think of it, I’m not even sure whether the tagging list is the right place to go with this. Anyway, I would love to hear any input, especially from the BiH community (whom I am also contacting by other means in this matter) and people who have been involved in any disputes regarding naming, disputed territories and the like.

To keep this concise and on-topic, I would like to stress that this proposal pertains strictly to the name=* tag. The usage of the language subtags (name:bs=*, name:hr=*, name:sr=*, name:sr-Latn=*) is well established in BiH and not a matter of dispute.

كتبت قبل عدة أيام سلاسل نصية وقواعد جديدة لاتجاهات صوتية في ملاحة OsmAnd.‏ للأسف, لا يوجد تصنيع صوتي Google لللغة العربية ولكن يمكنكم تنزيل أصوات عربية لبرنامجا تصنيع صوتي Vocalizer وA Capela.‏ يرجى اختبار وتصحيح من قبل متحدثين اللغة العربية.‏

Facebook grupa OpenStreetMap Bosna i Hercegovina

Posted by michalfabik on 28 January 2019 in Bosnian (Bosanski). Last updated on 31 January 2019.

Nakon nekoliko godina neaktivnosti javili su mi se osnivači Facebook grupe za OSM korisnike u BiH i dali su mi administratorska prava. Nemam iskustva sa administracijom na društvenim mrežama, ali ću se potruditi da se brinem za nju kako znam i umijem. Svratite, bacite like i ako imate bilo kakva pitanja ili sugestije što se tiče sadržaja, samo javite.

After several years of inactivity, the founders of the Facebook group for OSM users in Bosnia got back to me and gave me moderator rights. I have no previous experience with social network administration but I’ll try to take care of it as best I can. Visit us, like us, and if you have any questions or content suggestions, I’m all ears.

Hospitals in Bosnia

Posted by michalfabik on 12 February 2018 in English (English). Last updated on 13 February 2018.

The other day I noticed that there’s an absurdly large number of hospitals mapped in certain countries, including doctors’ offices, clinics, dentists’ offices, nursing homes, research institutes, pharmacies, even vets’ offices. Over the course of several days, I cleaned up Bosnia, leaving only true hospitals (i.e. provides inpatient care, consists of several specialized clinics, usually has the word “hospital” in the official name), and retagged everything else as appropriate. Here’s an image of hospitals in the former Yugoslavia with Bosnia cleaned up and the rest left untouched:

comparison of hospital density in Bosnia and the rest of Yugoslavia(full size)

Location: Donji Čajdraš, City of Zenica, Zenica-Doboj Canton, Federation of Bosnia and Herzegovina, 72260, Bosnia and Herzegovina

Promotivni flajeri za BiH

Posted by michalfabik on 25 December 2017 in Bosnian (Bosanski).

Evo stigli su:


Ko hoće da uzme za svoju školu, planinarsko društvo, lokalnu organizaciju izviđača ili bilo koje drugo mjesto gdje bi promocija OSM imala smisla, samo bujrum.

To my great delight, my girlfriend started mapping. In her case, that amounts to tracing missing roads and houses in backwoods villages (and bitching about overlapping CLC polygons) - she refuses to touch anything more complicated so far, in case she breaks it. The other day she commented how tracing stuff helps her relax a lot. I’d never thought about it that way but it kind of makes sense, I guess it’s similar to the adult colouring books that have been so popular lately.

Has anybody looked into this? Would it make sense to promote/teach OSM in places like nursing homes, hospitals, prisons etc., generally in places where people would appreciate an uncomplicated, quasi-creative activity to calm their nerves, keep their brains occupied or just kill time?

Balkan matters

Posted by michalfabik on 13 July 2017 in English (English).

Stuff like this is what I really hate about mapping in the Balkans. The memorial nearby is supposed to represent a mass grave and the tourist information name tag says “Mass grave of Bosniaks killed by the members of the HILLBILLYSTAN army!”, where “Hillbillystan” is a Bosnian mocking nickname of the so-called Serb Republic, the Serb-controlled territory within Bosnia and Herzegovina. Now I can either delete it and draw flak from Bosnian users, some of whom will certainly perceive it as an insult to their nation, or I can leave it alone and live with the fact that the map contains such utterly unprofessional childish garbage. (I actually contacted the author in a somewhat-related matter a while ago but got no answer.) And I’m certainly not helping matters by running into this just a couple of days after the Srebrenica massacre anniversary. Oh well.

EDIT: missed a word

Location: Vujasini, Novi Grad municipality, Republika Srpska, 79228, Bosnia and Herzegovina

Inspired by dval’s diary entry, I thought I should finally summarize me experience with track recording.

I use two setups on three physical devices to record my GPS tracks - my smarthphone and two old smartphones (identical models) that I’ve dedicated to GPS recording (taken out SIM cards, stripped the software down to bare essentials). My (newer, day-to-day use) smartphone is running the latest version of the Mendhak GPS Logger app, while the two older ones are running the most recent version of the same app that still supports Android 2.3.6 (i.e. a pretty old one). I have found that these two setups produce GPX tracks with different DOP values. The newer smartphone running the newer app usually produces tracks with DOP of around 0.4-5, the older smartphones with the older version of the app are a bit higher (DOP ~1.2-8). However, it doesn’t seem to depend entirely on the smartphone model because when I was still using the OsmAnd app for GPS tracking on the newer smartphone, the DOP values were a lot higher (~4-15 or thereabouts).

Every time I return from a trip or a hike, I download the GPX tracks from all three smartphones and in all of them I throw away all points with a negative elevation (<ele> tag) to get rid of complete rubbish. Then, for the newer smartphone, I throw away all poins with HDOP higher than 0.7 and for the older smartphones higher than 1.5. When I say “throw away”, I mean that I replace the track points in question (<trkpt> tag) with “</trkseg><trkseg>” so I don’t just connect the adjacent points. This way, I get a GPS track full of “holes” but since I’m recording three tracks at the same time and I often record the same trip on multiple occasions, I have a pretty decent coverage of the whole trip, as well as pretty good accuracy, I dare say.

Now, what do I do with this? I have 360 files so far and JOSM is starting to notice when I display them all at once. Of course, I’m thinking about uploading them to OSM because that way:
- other people could use them
- I would be able to only load the portions of tracks for a particular area, rather than needing to display them all in full in JOSM

- I don’t keep track of where I recorded a particular file and I certainly don’t feel like tagging them all for location etc. - the maintenance is tedious enough as it is.
- Many of them are next to useless on their own because they’re just a sparsely dispersed points or short dashes - it’s only after you combine more of them together that you see something meaningful.
- I don’t want to lose the HDOP information - the OSM server strips that, doesn’t it?
- I don’t want them to get lost in the vast number of other uploaded tracks, many of them low quality.

Basically, I’m looking for some place whence I could load them into JOSM along with tracks recorded by other people and then filter the lot by HDOP.

Any ideas?

While I was fixing a lot of stuff around Sarajevo, it occurred to me that we don’t have a way to render Muslim cemeteries - an important landmark of the cityscape there, and in many other parts of the world too, I’m sure. Some big cemeteries are even divided into Christian and Muslim sections.
Adapting one the existing rendering patterns was easy enough:
landuse=cemetery + religion=muslim
So, how does one officially propose a new way to render an existing feature?

Ako neko slučajno čita ovo, danas sam zatražio osnivanje mailing liste talk-ba. Uključite se i proširite ovu vijest.

Numbered shops

Posted by michalfabik on 19 April 2017 in English (English).

I’ve always been vaguely aware that individual branches within retail chains come with numbers. For example, my company’s business partners routinely refer to their branches as numbers, instead of locations: “Call five and ask how many of X they’ve got in stock”. Or: “Get me last month’s stats from three, eight and twelve”.
It wasn’t until a few days ago that I read a petrol station receipt I got (it said Petrol company so-and-so, branch 34) and it dawned on me that these numbers can be collected and entered into OSM using the “ref” tag. I’m not quite sure how exactly this can be useful to anyone but I do know for sure that people who work with data usually like their data numbered. So from now on, I’m shopping at 017, filling up my car at 34 and hunting for more receipts.

During a walk in a riparian forest, I happened upon a beaver dam:
beaver dam

and some construction works in progress:

beaver-gnawed tree

This got me thinking if it makes sense to map physical features created by animals. We’ve got the man_made tag, which some consider sexist and suggest that it be changed to human_made. Now what if I consider “human_made” too anthropocentric? Let’s use the more general animal_made=* + animal=human (or animal=beaver in this case).

OK, jokes aside. There are animal-made physical features which are reasonably permanent and prominent enough to serve for orientation. Beaver dams, stork nests, anthills and termite mounds come to mind. I’m sure there are more but I’m no biologist, maybe someone else will come with other examples.

Oh and BTW, how do you tag a riparian forest? It’s a forest with soaked ground and smelly pools of water here and there but nowhere near a swamp like in the film Southern Comfort (wetland=swamp).

Location: Nové Mlýny, Přítluky, okres Břeclav, Jihomoravský kraj, Southeast, Czechia

During our recent trip to Florence, a notoriously car-unfriendly city, I found this car park. I forgot to take a picture, but it obviously once was a paid car park that has been abandoned since. (By the authorities that is, not so much by the drivers.) The small building at the northern end looks like it might have served for the personnel and at the entrance and exit there are what looks like remains of lift gates that have been broken or sawn off. There are no blue signs with white “P” on them but there are no parking restrictions either. Many of the cars there were covered with a thick layer of dust, some had long (as in months) overdue parking tickets behind their wipers, others looked quite ordinary and came and went during our stay.
I asked some passers-by and all of them confirmed that it was ok to park there and that the locals regularly do so, only that we shouldn’t leave anything of value in the car. I left the car there and it was still there the next day, quite happy and unharmed.
Of course, I mapped the whole thing and proudly added a “fee=no”. The trouble is, technically it’s not an “amenity=parking” at all because it’s not designated as such on-site. “disused:amenity=parking” comes to mind but it’s not disused either, as people keep using it. “access=permissive” and possibly “access=discouraged” come close but not quite. So I added “informal=yes” like some footpaths have, although the car park certainly was “intentionally established” and didn’t just “evolve”, as the Wiki defines it. Actually, I’m pretty sure I know at least a few car parks where “informal=yes” would fit perfectly, but this isn’t one of them.
So, do we have a better way of dealing with this?

Location: Quartiere 3, Florence, Metropolitan City of Florence, Tuscany, Italy

Some time ago, I noticed that the Bosnian news portal sometimes use drone photos to accompany their news articles. I contacted the company and the author, their staff photographer Edin Hadžihasić, and they have kindly given their permission (in Serbo-Croatian) to use the pictures for tracing.

I aligned the photos with existing map data using Map Warper:

My workflow was as follows:
- pick control points recognizable in both Bing imagery and the drone pictures
- map the control points in OSM as accurately as possible (or adjust position of existing OSM objects)
- align the drone photo with the updated OSM map using Map Warper

Sure you can see how inefficient and error-prone this method is but I couldn’t find a way of aligning the drone photos with Bing directly, say by zooming and rotating a semi-opaque copy of the drone photo to fit with the corresponding part of Bing imagery. Any ideas?

Another problem is that there’s a lot of distortion in places with sloping terrain - this can be best seen in the picture with Žuta Tabija.

The results aren’t the most accurate thing in the world but as of now, there is no accurate imagery offset info available for the areas in question anyway, so I guess I didn’t break stuff more than it already was. After tracing the photos in JOSM, the results are quite pretty, I dare say:



There’s still stuff to be done of course. I plan on mapping driving lanes but I want to fix a couple of major intersections first and I need to do some on-site survey for that.

Location: Socijalno, Željeznička, Novo Sarajevo Municipality, City of Sarajevo, Sarajevo Canton, Federation of Bosnia and Herzegovina, 71000, Bosnia and Herzegovina